diff --git a/PPOCRLabel/README.md b/PPOCRLabel/README.md index c93c720f1e29bb18370475fd59c592d3eb985ebb..633d042b7308730816503470818d64982767a170 100644 --- a/PPOCRLabel/README.md +++ b/PPOCRLabel/README.md @@ -13,7 +13,7 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w - Undo function: In the process of drawing a four-point label box or after editing the box, press Ctrl+Z to undo the previous operation. - Fix image rotation and size problems, optimize the process of editing the mark frame (by [ninetailskim](https://github.com/ninetailskim)、 [edencfc](https://github.com/edencfc)). - 2021.1.11: Optimize the labeling experience (by [edencfc](https://github.com/edencfc)), - - Users can choose whether to pop up the label input dialog after drawing the detection box in "View - Pop-up Label Input Dialog". + - Users can choose whether to pop up the label input dialog after drawing the detection box in "View - Pop-up Label Input Dialog". - The recognition result scrolls synchronously when users click related detection box. - Click to modify the recognition result.(If you can't change the result, please switch to the system default input method, or switch back to the original input method again) - 2020.12.18: Support re-recognition of a single label box (by [ninetailskim](https://github.com/ninetailskim) ), perfect shortcut keys. @@ -52,7 +52,7 @@ python3 PPOCRLabel.py ``` pip3 install pyqt5 pip3 uninstall opencv-python # Uninstall opencv manually as it conflicts with pyqt -pip3 install opencv-contrib-python-headless # Install the headless version of opencv +pip3 install opencv-contrib-python-headless==4.2.0.32 # Install the headless version of opencv cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder python3 PPOCRLabel.py ``` @@ -129,7 +129,7 @@ python3 PPOCRLabel.py - Default model: PPOCRLabel uses the Chinese and English ultra-lightweight OCR model in PaddleOCR by default, supports Chinese, English and number recognition, and multiple language detection. -- Model language switching: Changing the built-in model language is supportable by clicking "PaddleOCR"-"Choose OCR Model" in the menu bar. Currently supported languages​include French, German, Korean, and Japanese. +- Model language switching: Changing the built-in model language is supportable by clicking "PaddleOCR"-"Choose OCR Model" in the menu bar. Currently supported languages​include French, German, Korean, and Japanese. For specific model download links, please refer to [PaddleOCR Model List](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md#multilingual-recognition-modelupdating) - Custom model: The model trained by users can be replaced by modifying PPOCRLabel.py in [PaddleOCR class instantiation](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/PPOCRLabel/PPOCRLabel.py#L110) referring [Custom Model Code](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/whl_en.md#use-custom-model) @@ -162,11 +162,11 @@ For some data that are difficult to recognize, the recognition results will not ``` pyrcc5 -o libs/resources.py resources.qrc ``` -- If you get an error ``` module 'cv2' has no attribute 'INTER_NEAREST'```, you need to delete all opencv related packages first, and then reinstall the headless version of opencv +- If you get an error ``` module 'cv2' has no attribute 'INTER_NEAREST'```, you need to delete all opencv related packages first, and then reinstall the 4.2.0.32 version of headless opencv ``` - pip install opencv-contrib-python-headless + pip install opencv-contrib-python-headless==4.2.0.32 ``` - + ### Related 1.[Tzutalin. LabelImg. Git code (2015)](https://github.com/tzutalin/labelImg) \ No newline at end of file diff --git a/PPOCRLabel/README_ch.md b/PPOCRLabel/README_ch.md index 7bc78800bef67f41f9d931a6bcd203bba8229abf..c214698f40fb8aac83de8bb06bdb1c8f9cac986e 100644 --- a/PPOCRLabel/README_ch.md +++ b/PPOCRLabel/README_ch.md @@ -52,7 +52,7 @@ python3 PPOCRLabel.py --lang ch ``` pip3 install pyqt5 pip3 uninstall opencv-python # 由于mac版本的opencv与pyqt有冲突，需先手动卸载opencv -pip3 install opencv-contrib-python-headless # 安装headless版本的open-cv +pip3 install opencv-contrib-python-headless==4.2.0.32 # 安装headless版本的open-cv cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下 python3 PPOCRLabel.py --lang ch ``` @@ -136,22 +136,22 @@ PPOCRLabel支持三种保存方式： ### 错误提示 - 如果同时使用whl包安装了paddleocr，其优先级大于通过paddleocr.py调用PaddleOCR类，whl包未更新时会导致程序异常。 - + - PPOCRLabel**不支持对中文文件名**的图片进行自动标注。 - 针对Linux用户：如果您在打开软件过程中出现**objc[XXXXX]**开头的错误，证明您的opencv版本太高，建议安装4.2版本： ``` pip install opencv-python==4.2.0.32 ``` - + - 如果出现 ```Missing string id``` 开头的错误，需要重新编译资源： ``` pyrcc5 -o libs/resources.py resources.qrc ``` - -- 如果出现``` module 'cv2' has no attribute 'INTER_NEAREST'```错误，需要首先删除所有opencv相关包，然后重新安装headless版本的opencv + +- 如果出现``` module 'cv2' has no attribute 'INTER_NEAREST'```错误，需要首先删除所有opencv相关包，然后重新安装4.2.0.32版本的headless opencv ``` - pip install opencv-contrib-python-headless + pip install opencv-contrib-python-headless==4.2.0.32 ``` ### 参考资料 diff --git a/README_ch.md b/README_ch.md index d627ec450f8260d1b88593ea14d9172d03016b5f..d4870710a24a3c17a997a603b34d447ce8f9f0bd 100755 --- a/README_ch.md +++ b/README_ch.md @@ -8,8 +8,8 @@ PaddleOCR同时支持动态图与静态图两种编程范式 - 静态图版本：develop分支 **近期更新** +- 2021.2.1 [FAQ](./doc/doc_ch/FAQ.md)新增5个高频问题，总数162个，每周一都会更新，欢迎大家持续关注。 - 2021.1.26,28,29 PaddleOCR官方研发团队带来技术深入解读三日直播课，1月26日、28日、29日晚上19:30，[直播地址](https://live.bilibili.com/21689802) -- 2021.1.25 [FAQ](./doc/doc_ch/FAQ.md)新增5个高频问题，总数157个，每周一都会更新，欢迎大家持续关注。 - 2021.1.21 更新多语言识别模型，目前支持语种超过27种，[多语言模型下载](./doc/doc_ch/models_list.md)，包括中文简体、中文繁体、英文、法文、德文、韩文、日文、意大利文、西班牙文、葡萄牙文、俄罗斯文、阿拉伯文等，后续计划可以参考[多语言研发计划](https://github.com/PaddlePaddle/PaddleOCR/issues/1048) - 2020.12.15 更新数据合成工具[Style-Text](./StyleText/README_ch.md)，可以批量合成大量与目标场景类似的图像，在多个场景验证，效果明显提升。 - 2020.11.25 更新半自动标注工具[PPOCRLabel](./PPOCRLabel/README_ch.md)，辅助开发者高效完成标注任务，输出格式与PP-OCR训练任务完美衔接。 diff --git a/StyleText/engine/predictors.py b/StyleText/engine/predictors.py index a1ba21f1b6cd084f9f95140d4227d600d4631715..ca9ab9ce6fc471e077970766c252c98b8617c6cc 100644 --- a/StyleText/engine/predictors.py +++ b/StyleText/engine/predictors.py @@ -38,7 +38,15 @@ class StyleTextRecPredictor(object): self.std = config["Predictor"]["std"] self.expand_result = config["Predictor"]["expand_result"] - def predict(self, style_input, text_input): + def reshape_to_same_height(self, img_list): + h = img_list[0].shape[0] + for idx in range(1, len(img_list)): + new_w = round(1.0 * img_list[idx].shape[1] / + img_list[idx].shape[0] * h) + img_list[idx] = cv2.resize(img_list[idx], (new_w, h)) + return img_list + + def predict_single_image(self, style_input, text_input): style_input = self.rep_style_input(style_input, text_input) tensor_style_input = self.preprocess(style_input) tensor_text_input = self.preprocess(text_input) @@ -64,6 +72,21 @@ class StyleTextRecPredictor(object): "fake_bg": fake_bg, } + def predict(self, style_input, text_input_list): + if not isinstance(text_input_list, (tuple, list)): + return self.predict_single_image(style_input, text_input_list) + + synth_result_list = [] + for text_input in text_input_list: + synth_result = self.predict_single_image(style_input, text_input) + synth_result_list.append(synth_result) + + for key in synth_result: + res = [r[key] for r in synth_result_list] + res = self.reshape_to_same_height(res) + synth_result[key] = np.concatenate(res, axis=1) + return synth_result + def preprocess(self, img): img = (img.astype('float32') * self.scale - self.mean) / self.std img_height, img_width, channel = img.shape diff --git a/StyleText/engine/synthesisers.py b/StyleText/engine/synthesisers.py index 177e3e049a695ecd06f5d2271f21336dd4eff997..6461d9e363f5f6e0c92831a50580c2748dffa248 100644 --- a/StyleText/engine/synthesisers.py +++ b/StyleText/engine/synthesisers.py @@ -12,6 +12,8 @@ # See the License for the specific language governing permissions and # limitations under the License. import os +import numpy as np +import cv2 from utils.config import ArgsParser, load_config, override_config from utils.logging import get_logger @@ -36,8 +38,9 @@ class ImageSynthesiser(object): self.predictor = getattr(predictors, predictor_method)(self.config) def synth_image(self, corpus, style_input, language="en"): - corpus, text_input = self.text_drawer.draw_text(corpus, language) - synth_result = self.predictor.predict(style_input, text_input) + corpus_list, text_input_list = self.text_drawer.draw_text( + corpus, language, style_input_width=style_input.shape[1]) + synth_result = self.predictor.predict(style_input, text_input_list) return synth_result @@ -59,12 +62,15 @@ class DatasetSynthesiser(ImageSynthesiser): for i in range(self.output_num): style_data = self.style_sampler.sample() style_input = style_data["image"] - corpus_language, text_input_label = self.corpus_generator.generate( - ) - text_input_label, text_input = self.text_drawer.draw_text( - text_input_label, corpus_language) + corpus_language, text_input_label = self.corpus_generator.generate() + text_input_label_list, text_input_list = self.text_drawer.draw_text( + text_input_label, + corpus_language, + style_input_width=style_input.shape[1]) - synth_result = self.predictor.predict(style_input, text_input) + text_input_label = "".join(text_input_label_list) + + synth_result = self.predictor.predict(style_input, text_input_list) fake_fusion = synth_result["fake_fusion"] self.writer.save_image(fake_fusion, text_input_label) self.writer.save_label() diff --git a/StyleText/engine/text_drawers.py b/StyleText/engine/text_drawers.py index 8aaac06ec50816bb6e2774972644c0a7dfb908c6..aeec75c3378f91b64b4387ef16971165f0b80ebe 100644 --- a/StyleText/engine/text_drawers.py +++ b/StyleText/engine/text_drawers.py @@ -1,5 +1,6 @@ from PIL import Image, ImageDraw, ImageFont import numpy as np +import cv2 from utils.logging import get_logger @@ -28,7 +29,11 @@ class StdTextDrawer(object): else: return int((self.height - 4)**2 / font_height) - def draw_text(self, corpus, language="en", crop=True): + def draw_text(self, + corpus, + language="en", + crop=True, + style_input_width=None): if language not in self.support_languages: self.logger.warning( "language {} not supported, use en instead.".format(language)) @@ -37,21 +42,43 @@ class StdTextDrawer(object): width = min(self.max_width, len(corpus) * self.height) + 4 else: width = len(corpus) * self.height + 4 - bg = Image.new("RGB", (width, self.height), color=(127, 127, 127)) - draw = ImageDraw.Draw(bg) - - char_x = 2 - font = self.font_dict[language] - for i, char_i in enumerate(corpus): - char_size = font.getsize(char_i)[0] - draw.text((char_x, 2), char_i, fill=(0, 0, 0), font=font) - char_x += char_size - if char_x >= width: - corpus = corpus[0:i + 1] - self.logger.warning("corpus length exceed limit: {}".format( - corpus)) + + if style_input_width is not None: + width = min(width, style_input_width) + + corpus_list = [] + text_input_list = [] + + while len(corpus) != 0: + bg = Image.new("RGB", (width, self.height), color=(127, 127, 127)) + draw = ImageDraw.Draw(bg) + char_x = 2 + font = self.font_dict[language] + i = 0 + while i < len(corpus): + char_i = corpus[i] + char_size = font.getsize(char_i)[0] + # split when char_x exceeds char size and index is not 0 (at least 1 char should be wroten on the image) + if char_x + char_size >= width and i != 0: + text_input = np.array(bg).astype(np.uint8) + text_input = text_input[:, 0:char_x, :] + + corpus_list.append(corpus[0:i]) + text_input_list.append(text_input) + corpus = corpus[i:] + break + draw.text((char_x, 2), char_i, fill=(0, 0, 0), font=font) + char_x += char_size + + i += 1 + # the whole text is shorter than style input + if i == len(corpus): + text_input = np.array(bg).astype(np.uint8) + text_input = text_input[:, 0:char_x, :] + + corpus_list.append(corpus[0:i]) + text_input_list.append(text_input) + corpus = corpus[i:] break - text_input = np.array(bg).astype(np.uint8) - text_input = text_input[:, 0:char_x, :] - return corpus, text_input + return corpus_list, text_input_list diff --git a/configs/cls/cls_mv3.yml b/configs/cls/cls_mv3.yml index b165bc4830f01f0e63c43b1d6a9635e432dc7605..5e643dc3839b2e2edf3c811db813dd6a90797366 100644 --- a/configs/cls/cls_mv3.yml +++ b/configs/cls/cls_mv3.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [0, 1000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: @@ -93,4 +92,4 @@ Eval: shuffle: False drop_last: False batch_size_per_card: 512 - num_workers: 4 \ No newline at end of file + num_workers: 4 diff --git a/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml b/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml index fd88495928b18c300386c6a9fd0cf57d840db21e..e001c376420c4210c69df90fa6564b1d903cf116 100644 --- a/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml +++ b/configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1200 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [3000, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained diff --git a/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml b/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml index 2694601254935be7d003148681334263d734579a..4229248df484d48a86d80c7362165f6a11acf32c 100644 --- a/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml +++ b/configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1200 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [3000, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/ResNet18_vd_pretrained diff --git a/configs/det/det_mv3_db.yml b/configs/det/det_mv3_db.yml index 00a16b5a38f2a271825a470684be401944eafec0..f8aab70543e15ca7e09d95cb3ea3da639692e170 100644 --- a/configs/det/det_mv3_db.yml +++ b/configs/det/det_mv3_db.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1200 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained diff --git a/configs/det/det_mv3_east.yml b/configs/det/det_mv3_east.yml index 05581a761cc91abce6d1e83bcdd7aacef1349148..187ac16054534da4f916c48dfb1b0e36441d331c 100644 --- a/configs/det/det_mv3_east.yml +++ b/configs/det/det_mv3_east.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1000 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [4000, 5000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained diff --git a/configs/det/det_r50_vd_db.yml b/configs/det/det_r50_vd_db.yml index 19c059d6737f9e98f33e6fc3b074587b24361dfc..3fa8948d17f1d84943c444035b8521e219e97209 100644 --- a/configs/det/det_r50_vd_db.yml +++ b/configs/det/det_r50_vd_db.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1200 # evaluation is run every 2000 iterations eval_batch_step: [0,2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained diff --git a/configs/det/det_r50_vd_east.yml b/configs/det/det_r50_vd_east.yml index b8fe55d4ac5473f6c0392820f5b1f651448bdddd..abef0b6116762fb2d5a1c8e7a797ed27832ffe55 100644 --- a/configs/det/det_r50_vd_east.yml +++ b/configs/det/det_r50_vd_east.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1000 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [4000, 5000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/ResNet50_vd_pretrained/ diff --git a/configs/det/det_r50_vd_sast_icdar15.yml b/configs/det/det_r50_vd_sast_icdar15.yml index a989bc8fc754ca88e3bff2de2a6db1060301fdd5..c24cae90132c68d662e9edb7a7975e358fb40d9c 100755 --- a/configs/det/det_r50_vd_sast_icdar15.yml +++ b/configs/det/det_r50_vd_sast_icdar15.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1000 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [4000, 5000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/ diff --git a/configs/det/det_r50_vd_sast_totaltext.yml b/configs/det/det_r50_vd_sast_totaltext.yml index e040c4207e497a7bf237a84c9c8d1b7c33a2dde8..e6f467c6ec78d453ce56fc0c9dffa35b71cb24c7 100755 --- a/configs/det/det_r50_vd_sast_totaltext.yml +++ b/configs/det/det_r50_vd_sast_totaltext.yml @@ -7,7 +7,10 @@ Global: save_epoch_step: 1000 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [4000, 5000] - # if pretrained_model is saved in static mode, load_static_weights must set to True + # 1. If pretrained_model is saved in static mode, such as classification pretrained model + # from static branch, load_static_weights must be set as True. + # 2. If you want to finetune the pretrained models we provide in the docs, + # you should set load_static_weights as False. load_static_weights: True cal_metric_during_train: False pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/ diff --git a/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml b/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml index 1db3e1cb8633f03f91d1d44064a19f7661e57b12..6a524e22cf4dea4c573d6b67e752c8527e973185 100644 --- a/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml +++ b/configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml b/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml index dc9d650f30f3d1086616a81c27aaf5db389a1fe7..c96621c5684f3861a7bc2f5aa8d9684e6512e228 100644 --- a/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml +++ b/configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_icdar15_train.yml b/configs/rec/rec_icdar15_train.yml index 8a743b501618dc6dabf9b1d419431a148236c690..5ae47c67d8b062746d422daac44011fb5aca38e2 100644 --- a/configs/rec/rec_icdar15_train.yml +++ b/configs/rec/rec_icdar15_train.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_mv3_none_bilstm_ctc.yml b/configs/rec/rec_mv3_none_bilstm_ctc.yml index 00c1db885e000d80ed3c3f42c2afbaa11c452ab5..900e98b6b34de824d9afb91b1867a86fe2debc24 100644 --- a/configs/rec/rec_mv3_none_bilstm_ctc.yml +++ b/configs/rec/rec_mv3_none_bilstm_ctc.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_mv3_none_none_ctc.yml b/configs/rec/rec_mv3_none_none_ctc.yml index 6711b1d23f843551d72e1dffc003637734727754..6d86b90c007cab9708cab6db6d8e3045dd5187fb 100644 --- a/configs/rec/rec_mv3_none_none_ctc.yml +++ b/configs/rec/rec_mv3_none_none_ctc.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_mv3_tps_bilstm_att.yml b/configs/rec/rec_mv3_tps_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..33aed74d83f9ab18b7bdfb12d0ed315c6eebc010 --- /dev/null +++ b/configs/rec/rec_mv3_tps_bilstm_att.yml @@ -0,0 +1,101 @@ +Global: + use_gpu: True + epoch_num: 72 + log_smooth_window: 20 + print_batch_step: 10 + save_model_dir: ./output/rec/rec_mv3_tps_bilstm_att/ + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: + character_type: en + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.0005 + regularizer: + name: 'L2' + factor: 0.00001 + +Architecture: + model_type: rec + algorithm: RARE + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: small + Backbone: + name: MobileNetV3 + scale: 0.5 + model_name: large + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 96 + Head: + name: AttentionHead + hidden_size: 96 + + +Loss: + name: AttentionLoss + +PostProcess: + name: AttnLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 1 diff --git a/configs/rec/rec_mv3_tps_bilstm_ctc.yml b/configs/rec/rec_mv3_tps_bilstm_ctc.yml index 1b9fb0a08db7cfd68bf2deb35d9216d68b58a12e..026c6a9dfbd6b6b543c0b4260c43cbf98e192e7b 100644 --- a/configs/rec/rec_mv3_tps_bilstm_ctc.yml +++ b/configs/rec/rec_mv3_tps_bilstm_ctc.yml @@ -1,5 +1,5 @@ Global: - use_gpu: true + use_gpu: True epoch_num: 72 log_smooth_window: 20 print_batch_step: 10 @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_r34_vd_none_bilstm_ctc.yml b/configs/rec/rec_r34_vd_none_bilstm_ctc.yml index e4d301a6a173ea772898c0528c4b3082670870ff..4052d426e51aa8c6e82ec216cfd65226922be602 100644 --- a/configs/rec/rec_r34_vd_none_bilstm_ctc.yml +++ b/configs/rec/rec_r34_vd_none_bilstm_ctc.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_r34_vd_none_none_ctc.yml b/configs/rec/rec_r34_vd_none_none_ctc.yml index 4a17a004228185db7e52dd71aadcff36d407d2cf..c3e1d9a3a91ab6a51e28d458623aea788b952ca0 100644 --- a/configs/rec/rec_r34_vd_none_none_ctc.yml +++ b/configs/rec/rec_r34_vd_none_none_ctc.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_r34_vd_tps_bilstm_att.yml b/configs/rec/rec_r34_vd_tps_bilstm_att.yml new file mode 100644 index 0000000000000000000000000000000000000000..87a14559849abd0a47c45f52b80d06bb0790ef0e --- /dev/null +++ b/configs/rec/rec_r34_vd_tps_bilstm_att.yml @@ -0,0 +1,100 @@ +Global: + use_gpu: True + epoch_num: 400 + log_smooth_window: 20 + print_batch_step: 10 + save_model_dir: ./output/rec/b3_rare_r34_none_gru/ + save_epoch_step: 3 + # evaluation is run every 5000 iterations after the 4000th iteration + eval_batch_step: [0, 2000] + cal_metric_during_train: True + pretrained_model: + checkpoints: + save_inference_dir: + use_visualdl: False + infer_img: doc/imgs_words/ch/word_1.jpg + # for data or label process + character_dict_path: + character_type: en + max_text_length: 25 + infer_mode: False + use_space_char: False + + +Optimizer: + name: Adam + beta1: 0.9 + beta2: 0.999 + lr: + learning_rate: 0.0005 + regularizer: + name: 'L2' + factor: 0.00000 + +Architecture: + model_type: rec + algorithm: RARE + Transform: + name: TPS + num_fiducial: 20 + loc_lr: 0.1 + model_name: large + Backbone: + name: ResNet + layers: 34 + Neck: + name: SequenceEncoder + encoder_type: rnn + hidden_size: 256 #96 + Head: + name: AttentionHead # AttentionHead + hidden_size: 256 # + l2_decay: 0.00001 + +Loss: + name: AttentionLoss + +PostProcess: + name: AttnLabelDecode + +Metric: + name: RecMetric + main_indicator: acc + +Train: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/training/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: True + batch_size_per_card: 256 + drop_last: True + num_workers: 8 + +Eval: + dataset: + name: LMDBDataSet + data_dir: ./train_data/data_lmdb_release/validation/ + transforms: + - DecodeImage: # load image + img_mode: BGR + channel_first: False + - AttnLabelEncode: # Class handling label + - RecResizeImg: + image_shape: [3, 32, 100] + - KeepKeys: + keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order + loader: + shuffle: False + drop_last: False + batch_size_per_card: 256 + num_workers: 8 diff --git a/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml b/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml index 62edf84379ec1be9ef5f7155b240099f5fbb7b00..8b80b82dd6e16f4b4a921f7a36531755852bd70a 100644 --- a/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml +++ b/configs/rec/rec_r34_vd_tps_bilstm_ctc.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 2000 iterations eval_batch_step: [0, 2000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: diff --git a/configs/rec/rec_r50_fpn_srn.yml b/configs/rec/rec_r50_fpn_srn.yml index ec7f170560f5309818d537953a93c180b9de0bb7..34a997f3c3f6bc3dccdc62757c6adb1e1d17cc53 100644 --- a/configs/rec/rec_r50_fpn_srn.yml +++ b/configs/rec/rec_r50_fpn_srn.yml @@ -7,7 +7,6 @@ Global: save_epoch_step: 3 # evaluation is run every 5000 iterations after the 4000th iteration eval_batch_step: [0, 5000] - # if pretrained_model is saved in static mode, load_static_weights must set to True cal_metric_during_train: True pretrained_model: checkpoints: @@ -59,7 +58,7 @@ Metric: Train: dataset: name: LMDBDataSet - data_dir: ./train_data/srn_train_data_duiqi + data_dir: ./train_data/data_lmdb_release/training/ transforms: - DecodeImage: # load image img_mode: BGR @@ -84,7 +83,7 @@ Train: Eval: dataset: name: LMDBDataSet - data_dir: ./train_data/data_lmdb_release/evaluation + data_dir: ./train_data/data_lmdb_release/validation/ transforms: - DecodeImage: # load image img_mode: BGR diff --git a/deploy/cpp_infer/CMakeLists.txt b/deploy/cpp_infer/CMakeLists.txt index 1188336730ea9fbaac11a84250a3be3e418ec5f5..120cf06a04c8f1769bdcd153a74cd6ff0665cf93 100644 --- a/deploy/cpp_infer/CMakeLists.txt +++ b/deploy/cpp_infer/CMakeLists.txt @@ -133,7 +133,11 @@ if(WITH_MKL) endif () endif() else() - set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/libopenblas${CMAKE_STATIC_LIBRARY_SUFFIX}) + if (WIN32) + set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/openblas${CMAKE_STATIC_LIBRARY_SUFFIX}) + else () + set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/libopenblas${CMAKE_STATIC_LIBRARY_SUFFIX}) + endif () endif() # Note: libpaddle_inference_api.so/a must put before libpaddle_fluid.so/a @@ -157,7 +161,7 @@ endif(WITH_STATIC_LIB) if (NOT WIN32) set(DEPS ${DEPS} - ${MATH_LIB} ${MKLDNN_LIB} + ${MATH_LIB} ${MKLDNN_LIB} glog gflags protobuf z xxhash ) if(EXISTS "${PADDLE_LIB}/third_party/install/snappystream/lib") diff --git a/deploy/cpp_infer/docs/windows_vs2019_build.md b/deploy/cpp_infer/docs/windows_vs2019_build.md index 21fbf4e0eb95ee82475164047d8051e90e9e224f..0f243bf8f54b5cd50e9fa2faab29b064e694e45c 100644 --- a/deploy/cpp_infer/docs/windows_vs2019_build.md +++ b/deploy/cpp_infer/docs/windows_vs2019_build.md @@ -14,7 +14,7 @@ PaddleOCR在Windows 平台下基于`Visual Studio 2019 Community` 进行了测 ### Step1: 下载PaddlePaddle C++ 预测库 fluid_inference -PaddlePaddle C++ 预测库针对不同的`CPU`和`CUDA`版本提供了不同的预编译版本，请根据实际情况下载: [C++预测库下载列表](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/advanced_guide/inference_deployment/inference/windows_cpp_inference.html) +PaddlePaddle C++ 预测库针对不同的`CPU`和`CUDA`版本提供了不同的预编译版本，请根据实际情况下载: [C++预测库下载列表](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/windows_cpp_inference.html) 解压后`D:\projects\fluid_inference`目录包含内容为： ``` diff --git a/deploy/cpp_infer/readme.md b/deploy/cpp_infer/readme.md index b563ecf48c2aba03e25a03ae0328c244bb900356..633553082160ddc37944feadd35287ebcee9bdef 100644 --- a/deploy/cpp_infer/readme.md +++ b/deploy/cpp_infer/readme.md @@ -1,6 +1,8 @@ # 服务器端C++预测 -本教程将介绍在服务器端部署PaddleOCR超轻量中文检测、识别模型的详细步骤。 +本章节介绍PaddleOCR 模型的的C++部署方法，与之对应的python预测部署方式参考[文档](../../doc/doc_ch/inference.md)。 +C++在性能计算上优于python，因此，在大多数CPU、GPU部署场景，多采用C++的部署方式，本节将介绍如何在Linux\Windows (CPU\GPU)环境下配置C++环境并完成 +PaddleOCR模型部署。 ## 1. 准备环境 @@ -72,9 +74,21 @@ opencv3/ * 有2种方式获取Paddle预测库，下面进行详细介绍。 -#### 1.2.1 预测库源码编译 +#### 1.2.1 直接下载安装 + +* [Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)上提供了不同cuda版本的Linux预测库，可以在官网查看并选择合适的预测库版本。 + +* 下载之后使用下面的方法解压。 + +``` +tar -xf paddle_inference.tgz +``` + +最终会在当前的文件夹中生成`paddle_inference/`的子文件夹。 + +#### 1.2.2 预测库源码编译 * 如果希望获取最新预测库特性，可以从Paddle github上克隆最新代码，源码编译预测库。 -* 可以参考[Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)的说明，从github上获取Paddle代码，然后进行编译，生成最新的预测库。使用git获取代码方法如下。 +* 可以参考[Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)的说明，从github上获取Paddle代码，然后进行编译，生成最新的预测库。使用git获取代码方法如下。 ```shell git clone https://github.com/PaddlePaddle/Paddle.git @@ -100,7 +114,7 @@ make -j make inference_lib_dist ``` -更多编译参数选项可以参考Paddle C++预测库官网：[https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)。 +更多编译参数选项可以参考Paddle C++预测库官网：[https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html](https://www.paddlepaddle.org.cn/documentation/docs/zh/develop/guides/05_inference_deployment/inference/build_and_install_lib_cn.html)。 * 编译完成之后，可以在`build/paddle_inference_install_dir/`文件下看到生成了以下文件及文件夹。 @@ -115,17 +129,7 @@ build/paddle_inference_install_dir/ 其中`paddle`就是C++预测所需的Paddle库，`version.txt`中包含当前预测库的版本信息。 -#### 1.2.2 直接下载安装 - -* [Paddle预测库官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html)上提供了不同cuda版本的Linux预测库，可以在官网查看并选择合适的预测库版本。 -* 下载之后使用下面的方法解压。 - -``` -tar -xf paddle_inference.tgz -``` - -最终会在当前的文件夹中生成`paddle_inference/`的子文件夹。 ## 2 开始运行 @@ -223,7 +227,7 @@ char_list_file ../../ppocr/utils/ppocr_keys_v1.txt # 字典文件 visualize 1 # 是否对结果进行可视化，为1时，会在当前文件夹下保存文件名为`ocr_vis.png`的预测结果。 ``` -* PaddleOCR也支持多语言的预测，更多细节可以参考[识别文档](../../doc/doc_ch/recognition.md)中的多语言字典与模型部分。 +* PaddleOCR也支持多语言的预测，更多支持的语言和模型可以参考[识别文档](../../doc/doc_ch/recognition.md)中的多语言字典与模型部分，如果希望进行多语言预测，只需将修改`tools/config.txt`中的`char_list_file`（字典文件路径）以及`rec_model_dir`（inference模型路径）字段即可。 最终屏幕上会输出检测结果如下。 @@ -234,4 +238,4 @@ visualize 1 # 是否对结果进行可视化，为1时，会在当前文件夹 ### 2.3 注意 -* 在使用Paddle预测库时，推荐使用2.0.0-beta0版本的预测库。 +* 在使用Paddle预测库时，推荐使用2.0.0版本的预测库。 diff --git a/deploy/cpp_infer/readme_en.md b/deploy/cpp_infer/readme_en.md index 41c764bc18a69965da6ad2ea521f438840c286e6..cfe18a286fef022d83e61ea02465c0497f0f3d8c 100644 --- a/deploy/cpp_infer/readme_en.md +++ b/deploy/cpp_infer/readme_en.md @@ -1,7 +1,9 @@ # Server-side C++ inference - -In this tutorial, we will introduce the detailed steps of deploying PaddleOCR ultra-lightweight Chinese detection and recognition models on the server side. +This chapter introduces the C++ deployment method of the PaddleOCR model, and the corresponding python predictive deployment method refers to [document](../../doc/doc_ch/inference.md). +C++ is better than python in terms of performance calculation. Therefore, in most CPU and GPU deployment scenarios, C++ deployment is mostly used. +This section will introduce how to configure the C++ environment and complete it in the Linux\Windows (CPU\GPU) environment +PaddleOCR model deployment. ## 1. Prepare the environment @@ -74,10 +76,23 @@ opencv3/ * There are 2 ways to obtain the Paddle inference library, described in detail below. +#### 1.2.1 Direct download and installation + +* Different cuda versions of the Linux inference library (based on GCC 4.8.2) are provided on the +[Paddle inference library official website](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html). You can view and select the appropriate version of the inference library on the official website. + -#### 1.2.1 Compile from the source code +* After downloading, use the following method to uncompress. + +``` +tar -xf paddle_inference.tgz +``` + +Finally you can see the following files in the folder of `paddle_inference/`. + +#### 1.2.2 Compile from the source code * If you want to get the latest Paddle inference library features, you can download the latest code from Paddle github repository and compile the inference library from the source code. -* You can refer to [Paddle inference library] (https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html) to get the Paddle source code from github, and then compile To generate the latest inference library. The method of using git to access the code is as follows. +* You can refer to [Paddle inference library] (https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html) to get the Paddle source code from github, and then compile To generate the latest inference library. The method of using git to access the code is as follows. ```shell @@ -104,7 +119,7 @@ make -j make inference_lib_dist ``` -For more compilation parameter options, please refer to the official website of the Paddle C++ inference library:[https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html](https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html). +For more compilation parameter options, please refer to the official website of the Paddle C++ inference library:[https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html](https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/05_inference_deployment/inference/build_and_install_lib_en.html). * After the compilation process, you can see the following files in the folder of `build/paddle_inference_install_dir/`. @@ -120,22 +135,6 @@ build/paddle_inference_install_dir/ Among them, `paddle` is the Paddle library required for C++ prediction later, and `version.txt` contains the version information of the current inference library. - -#### 1.2.2 Direct download and installation - -* Different cuda versions of the Linux inference library (based on GCC 4.8.2) are provided on the -[Paddle inference library official website](https://www.paddlepaddle.org.cn/documentation/docs/en/advanced_guide/inference_deployment/inference/build_and_install_lib_en.html). You can view and select the appropriate version of the inference library on the official website. - - -* After downloading, use the following method to uncompress. - -``` -tar -xf paddle_inference.tgz -``` - -Finally you can see the following files in the folder of `paddle_inference/`. - - ## 2. Compile and run the demo ### 2.1 Export the inference model @@ -233,7 +232,7 @@ char_list_file ../../ppocr/utils/ppocr_keys_v1.txt # dictionary file visualize 1 # Whether to visualize the results，when it is set as 1, The prediction result will be save in the image file `./ocr_vis.png`. ``` -* Multi-language inference is also supported in PaddleOCR, for more details, please refer to part of multi-language dictionaries and models in [recognition tutorial](../../doc/doc_en/recognition_en.md). +* Multi-language inference is also supported in PaddleOCR, you can refer to [recognition tutorial](../../doc/doc_en/recognition_en.md) for more supported languages and models in PaddleOCR. Specifically, if you want to infer using multi-language models, you just need to modify values of `char_list_file` and `rec_model_dir` in file `tools/config.txt`. The detection results will be shown on the screen, which is as follows. @@ -245,4 +244,4 @@ The detection results will be shown on the screen, which is as follows. ### 2.3 Notes -* Paddle2.0.0-beta0 inference model library is recommended for this toturial. +* Paddle2.0.0 inference model library is recommended for this toturial. diff --git a/deploy/cpp_infer/tools/config.txt b/deploy/cpp_infer/tools/config.txt index e185377e2f2c9cbd5c1d8ed09ba43df9c41c05d2..24e4ef0de7d844ba4bd6c11f2cba08766c0e5ddf 100644 --- a/deploy/cpp_infer/tools/config.txt +++ b/deploy/cpp_infer/tools/config.txt @@ -9,7 +9,7 @@ use_mkldnn 0 max_side_len 960 det_db_thresh 0.3 det_db_box_thresh 0.5 -det_db_unclip_ratio 2.0 +det_db_unclip_ratio 1.6 det_model_dir ./inference/ch_ppocr_mobile_v2.0_det_infer/ # cls config diff --git a/deploy/docker/hubserving/cpu/Dockerfile b/deploy/docker/hubserving/cpu/Dockerfile index e46ca73be0a894631fe7e11cda3bdf7a20ad5526..ef1a7b7b16fe1ebe26b5681a72f64d404dd77608 100644 --- a/deploy/docker/hubserving/cpu/Dockerfile +++ b/deploy/docker/hubserving/cpu/Dockerfile @@ -1,5 +1,5 @@ # Version: 2.0.0 -FROM registry.baidubce.com/paddlepaddle/paddle:2.0.0rc1 +FROM registry.baidubce.com/paddlepaddle/paddle:2.0.0 # PaddleOCR base on Python3.7 RUN pip3.7 install --upgrade pip -i https://mirror.baidu.com/pypi/simple diff --git a/deploy/docker/hubserving/gpu/Dockerfile b/deploy/docker/hubserving/gpu/Dockerfile index b7fa6f4ca0467b43186073c82c9ac5afc4b1216e..b3f2d21ffec340a93894f089d72ae775c428ef33 100644 --- a/deploy/docker/hubserving/gpu/Dockerfile +++ b/deploy/docker/hubserving/gpu/Dockerfile @@ -1,5 +1,5 @@ # Version: 2.0.0 -FROM egistry.baidubce.com/paddlepaddle/paddle:2.0.0rc1-gpu-cuda10.0-cudnn7 +FROM registry.baidubce.com/paddlepaddle/paddle:2.0.0-gpu-cuda10.1-cudnn7 # PaddleOCR base on Python3.7 RUN pip3.7 install --upgrade pip -i https://mirror.baidu.com/pypi/simple diff --git a/deploy/hubserving/ocr_det/params.py b/deploy/hubserving/ocr_det/params.py index 132158904d44a5a45600e6cfc9cd3e565ddcef0b..7be88e9bc6673fadfce19a281de4ba4d2b235fd2 100755 --- a/deploy/hubserving/ocr_det/params.py +++ b/deploy/hubserving/ocr_det/params.py @@ -20,7 +20,8 @@ def read_params(): #DB parmas cfg.det_db_thresh = 0.3 cfg.det_db_box_thresh = 0.5 - cfg.det_db_unclip_ratio = 2.0 + cfg.det_db_unclip_ratio = 1.6 + cfg.use_dilation = False # #EAST parmas # cfg.det_east_score_thresh = 0.8 diff --git a/deploy/hubserving/ocr_system/params.py b/deploy/hubserving/ocr_system/params.py index add466668eee0be1e1674fce5f5a07c24c0c5e3f..bd56dc2e8fc05309e27227d25975dc784a17c2cf 100755 --- a/deploy/hubserving/ocr_system/params.py +++ b/deploy/hubserving/ocr_system/params.py @@ -20,7 +20,8 @@ def read_params(): #DB parmas cfg.det_db_thresh = 0.3 cfg.det_db_box_thresh = 0.5 - cfg.det_db_unclip_ratio = 2.0 + cfg.det_db_unclip_ratio = 1.6 + cfg.use_dilation = False #EAST parmas cfg.det_east_score_thresh = 0.8 diff --git a/deploy/slim/prune/README.md b/deploy/slim/prune/README.md new file mode 100644 index 0000000000000000000000000000000000000000..d9675c5a3cfc281a3a2af69b364bd528597096d7 --- /dev/null +++ b/deploy/slim/prune/README.md @@ -0,0 +1,64 @@ + +## 介绍 + +复杂的模型有利于提高模型的性能，但也导致模型中存在一定冗余，模型裁剪通过移出网络模型中的子模型来减少这种冗余，达到减少模型计算复杂度，提高模型推理性能的目的。 +本教程将介绍如何使用飞桨模型压缩库PaddleSlim做PaddleOCR模型的压缩。 +[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)集成了模型剪枝、量化（包括量化训练和离线量化）、蒸馏和神经网络搜索等多种业界常用且领先的模型压缩功能，如果您感兴趣，可以关注并了解。 + + +在开始本教程之前，建议先了解： +1. [PaddleOCR模型的训练方法](../../../doc/doc_ch/quickstart.md) +2. [模型裁剪教程](https://github.com/PaddlePaddle/PaddleSlim/blob/release%2F2.0.0/docs/zh_cn/tutorials/pruning/dygraph/filter_pruning.md) + + +## 快速开始 + +模型裁剪主要包括四个步骤： +1. 安装 PaddleSlim +2. 准备训练好的模型 +3. 敏感度分析、裁剪训练 +4. 导出模型、预测部署 + +### 1. 安装PaddleSlim + +```bash +git clone https://github.com/PaddlePaddle/PaddleSlim.git +git checkout develop +cd Paddleslim +python3 setup.py install +``` + +### 2. 获取预训练模型 +模型裁剪需要加载事先训练好的模型，PaddleOCR也提供了一系列(模型)[../../../doc/doc_ch/models_list.md]，开发者可根据需要自行选择模型或使用自己的模型。 + +### 3. 敏感度分析训练 + +加载预训练模型后，通过对现有模型的每个网络层进行敏感度分析，得到敏感度文件：sen.pickle，可以通过PaddleSlim提供的[接口](https://github.com/PaddlePaddle/PaddleSlim/blob/9b01b195f0c4bc34a1ab434751cb260e13d64d9e/paddleslim/dygraph/prune/filter_pruner.py#L75)加载文件，获得各网络层在不同裁剪比例下的精度损失。从而了解各网络层冗余度，决定每个网络层的裁剪比例。 +敏感度文件内容格式： + sen.pickle(Dict){ + 'layer_weight_name_0': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss} + 'layer_weight_name_1': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss} + } + + 例子： + { + 'conv10_expand_weights': {0.1: 0.006509952684312718, 0.2: 0.01827734339798862, 0.3: 0.014528405644659832, 0.6: 0.06536008804270439, 0.8: 0.11798612250664964, 0.7: 0.12391408417493704, 0.4: 0.030615754498018757, 0.5: 0.047105205602406594} + 'conv10_linear_weights': {0.1: 0.05113190831455035, 0.2: 0.07705573833558801, 0.3: 0.12096721757739311, 0.6: 0.5135061352930738, 0.8: 0.7908166677143281, 0.7: 0.7272187676899062, 0.4: 0.1819252083008504, 0.5: 0.3728054727792405} + } +加载敏感度文件后会返回一个字典，字典中的keys为网络模型参数模型的名字，values为一个字典，里面保存了相应网络层的裁剪敏感度信息。例如在例子中，conv10_expand_weights所对应的网络层在裁掉10%的卷积核后模型性能相较原模型会下降0.65%，详细信息可见[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/algo/algo.md#2-%E5%8D%B7%E7%A7%AF%E6%A0%B8%E5%89%AA%E8%A3%81%E5%8E%9F%E7%90%86) + +进入PaddleOCR根目录，通过以下命令对模型进行敏感度分析训练： +```bash +python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrain_weights="your trained model" +``` + +### 4. 导出模型、预测部署 + +在得到裁剪训练保存的模型后，我们可以将其导出为inference_model： +```bash +pytho3.7 deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrain_weights=./output/det_db/best_accuracy Global.save_inference_dir=inference_model +``` + +inference model的预测和部署参考： +1. [inference model python端预测](../../../doc/doc_ch/inference.md) +2. [inference model C++预测](../../cpp_infer/readme.md) diff --git a/deploy/slim/prune/README_en.md b/deploy/slim/prune/README_en.md new file mode 100644 index 0000000000000000000000000000000000000000..70cfd580b30dde2070e27cd3512f54f222acfaed --- /dev/null +++ b/deploy/slim/prune/README_en.md @@ -0,0 +1,71 @@ + +## Introduction + +Generally, a more complex model would achive better performance in the task, but it also leads to some redundancy in the model. Model Pruning is a technique that reduces this redundancy by removing the sub-models in the neural network model, so as to reduce model calculation complexity and improve model inference performance. + +This example uses PaddleSlim provided[APIs of Pruning](https://paddlepaddle.github.io/PaddleSlim/api/prune_api/) to compress the OCR model. +[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim), an open source library which integrates model pruning, quantization (including quantization training and offline quantization), distillation, neural network architecture search, and many other commonly used and leading model compression technique in the industry. + +It is recommended that you could understand following pages before reading this example： +1. [PaddleOCR training methods](../../../doc/doc_ch/quickstart.md) +2. [The demo of prune](https://github.com/PaddlePaddle/PaddleSlim/blob/release%2F2.0.0/docs/zh_cn/tutorials/pruning/dygraph/filter_pruning.md) + +## Quick start + +Five steps for OCR model prune: +1. Install PaddleSlim +2. Prepare the trained model +3. Sensitivity analysis and tailoring training +4. Export model, predict deployment + +### 1. Install PaddleSlim + +```bash +git clone https://github.com/PaddlePaddle/PaddleSlim.git +git checkout develop +cd Paddleslim +python3 setup.py install +``` + + +### 2. Download Pretrain Model +Model prune needs to load pre-trained models. +PaddleOCR also provides a series of (models)[../../../doc/doc_en/models_list_en.md]. Developers can choose their own models or use their own models according to their needs. + + +### 3. Pruning sensitivity analysis + + After the pre-training model is loaded, sensitivity analysis is performed on each network layer of the model to understand the redundancy of each network layer, and save a sensitivity file which named: sen.pickle. After that, user could load the sensitivity file via the [methods provided by PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/paddleslim/prune/sensitive.py#L221) and determining the pruning ratio of each network layer automatically. For specific details of sensitivity analysis, see：[Sensitivity analysis](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/tutorials/image_classification_sensitivity_analysis_tutorial.md) + The data format of sensitivity file： + sen.pickle(Dict){ + 'layer_weight_name_0': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss} + 'layer_weight_name_1': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss} + } + + example： + { + 'conv10_expand_weights': {0.1: 0.006509952684312718, 0.2: 0.01827734339798862, 0.3: 0.014528405644659832, 0.6: 0.06536008804270439, 0.8: 0.11798612250664964, 0.7: 0.12391408417493704, 0.4: 0.030615754498018757, 0.5: 0.047105205602406594} + 'conv10_linear_weights': {0.1: 0.05113190831455035, 0.2: 0.07705573833558801, 0.3: 0.12096721757739311, 0.6: 0.5135061352930738, 0.8: 0.7908166677143281, 0.7: 0.7272187676899062, 0.4: 0.1819252083008504, 0.5: 0.3728054727792405} + } + The function would return a dict after loading the sensitivity file. The keys of the dict are name of parameters in each layer. And the value of key is the information about pruning sensitivity of correspoding layer. In example, pruning 10% filter of the layer corresponding to conv10_expand_weights would lead to 0.65% degradation of model performance. The details could be seen at: [Sensitivity analysis](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/algo/algo.md#2-%E5%8D%B7%E7%A7%AF%E6%A0%B8%E5%89%AA%E8%A3%81%E5%8E%9F%E7%90%86) + + +Enter the PaddleOCR root directory，perform sensitivity analysis on the model with the following command： + +```bash + +python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrain_weights="your trained model" + +``` + + +### 5. Export inference model and deploy it + +We can export the pruned model as inference_model for deployment: +```bash +python deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrain_weights=./output/det_db/best_accuracy Global.test_batch_size_per_card=1 Global.save_inference_dir=inference_model +``` + +Reference for prediction and deployment of inference model: +1. [inference model python prediction](../../../doc/doc_en/inference_en.md) +2. [inference model C++ prediction](../../cpp_infer/readme_en.md) diff --git a/deploy/slim/prune/export_prune_model.py b/deploy/slim/prune/export_prune_model.py new file mode 100644 index 0000000000000000000000000000000000000000..29f7d211df7b2ad02bf2229f0be81c3cbe005503 --- /dev/null +++ b/deploy/slim/prune/export_prune_model.py @@ -0,0 +1,125 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys + +__dir__ = os.path.dirname(__file__) +sys.path.append(__dir__) +sys.path.append(os.path.join(__dir__, '..', '..', '..')) +sys.path.append(os.path.join(__dir__, '..', '..', '..', 'tools')) + +import paddle +from ppocr.data import build_dataloader +from ppocr.modeling.architectures import build_model + +from ppocr.postprocess import build_post_process +from ppocr.metrics import build_metric +from ppocr.utils.save_load import init_model +import tools.program as program + + +def main(config, device, logger, vdl_writer): + + global_config = config['Global'] + + # build dataloader + valid_dataloader = build_dataloader(config, 'Eval', device, logger) + + # build post process + post_process_class = build_post_process(config['PostProcess'], + global_config) + + # build model + # for rec algorithm + if hasattr(post_process_class, 'character'): + char_num = len(getattr(post_process_class, 'character')) + config['Architecture']["Head"]['out_channels'] = char_num + model = build_model(config['Architecture']) + + flops = paddle.flops(model, [1, 3, 640, 640]) + logger.info(f"FLOPs before pruning: {flops}") + + from paddleslim.dygraph import FPGMFilterPruner + model.train() + pruner = FPGMFilterPruner(model, [1, 3, 640, 640]) + + # build metric + eval_class = build_metric(config['Metric']) + + def eval_fn(): + metric = program.eval(model, valid_dataloader, post_process_class, + eval_class) + logger.info(f"metric['hmean']: {metric['hmean']}") + return metric['hmean'] + + params_sensitive = pruner.sensitive( + eval_func=eval_fn, + sen_file="./sen.pickle", + skip_vars=[ + "conv2d_57.w_0", "conv2d_transpose_2.w_0", "conv2d_transpose_3.w_0" + ]) + + logger.info( + "The sensitivity analysis results of model parameters saved in sen.pickle" + ) + # calculate pruned params's ratio + params_sensitive = pruner._get_ratios_by_loss(params_sensitive, loss=0.02) + for key in params_sensitive.keys(): + logger.info(f"{key}, {params_sensitive[key]}") + + plan = pruner.prune_vars(params_sensitive, [0]) + + flops = paddle.flops(model, [1, 3, 640, 640]) + logger.info(f"FLOPs after pruning: {flops}") + + # load pretrain model + pre_best_model_dict = init_model(config, model, logger, None) + metric = program.eval(model, valid_dataloader, post_process_class, + eval_class) + logger.info(f"metric['hmean']: {metric['hmean']}") + + # start export model + from paddle.jit import to_static + + infer_shape = [3, -1, -1] + if config['Architecture']['model_type'] == "rec": + infer_shape = [3, 32, -1] # for rec model, H must be 32 + + if 'Transform' in config['Architecture'] and config['Architecture'][ + 'Transform'] is not None and config['Architecture'][ + 'Transform']['name'] == 'TPS': + logger.info( + 'When there is tps in the network, variable length input is not supported, and the input size needs to be the same as during training' + ) + infer_shape[-1] = 100 + model = to_static( + model, + input_spec=[ + paddle.static.InputSpec( + shape=[None] + infer_shape, dtype='float32') + ]) + + save_path = '{}/inference'.format(config['Global']['save_inference_dir']) + paddle.jit.save(model, save_path) + logger.info('inference model is saved to {}'.format(save_path)) + + +if __name__ == '__main__': + config, device, logger, vdl_writer = program.preprocess(is_train=True) + main(config, device, logger, vdl_writer) diff --git a/deploy/slim/prune/sensitivity_anal.py b/deploy/slim/prune/sensitivity_anal.py new file mode 100644 index 0000000000000000000000000000000000000000..bd2b96497221fd886c83b9401cc8ed2a1a201a50 --- /dev/null +++ b/deploy/slim/prune/sensitivity_anal.py @@ -0,0 +1,146 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys + +__dir__ = os.path.dirname(__file__) +sys.path.append(__dir__) +sys.path.append(os.path.join(__dir__, '..', '..', '..')) +sys.path.append(os.path.join(__dir__, '..', '..', '..', 'tools')) + +import paddle +import paddle.distributed as dist +from ppocr.data import build_dataloader +from ppocr.modeling.architectures import build_model +from ppocr.losses import build_loss +from ppocr.optimizer import build_optimizer +from ppocr.postprocess import build_post_process +from ppocr.metrics import build_metric +from ppocr.utils.save_load import init_model +import tools.program as program + +dist.get_world_size() + + +def get_pruned_params(parameters): + params = [] + + for param in parameters: + if len( + param.shape + ) == 4 and 'depthwise' not in param.name and 'transpose' not in param.name and "conv2d_57" not in param.name and "conv2d_56" not in param.name: + params.append(param.name) + return params + + +def main(config, device, logger, vdl_writer): + # init dist environment + if config['Global']['distributed']: + dist.init_parallel_env() + + global_config = config['Global'] + + # build dataloader + train_dataloader = build_dataloader(config, 'Train', device, logger) + if config['Eval']: + valid_dataloader = build_dataloader(config, 'Eval', device, logger) + else: + valid_dataloader = None + + # build post process + post_process_class = build_post_process(config['PostProcess'], + global_config) + + # build model + # for rec algorithm + if hasattr(post_process_class, 'character'): + char_num = len(getattr(post_process_class, 'character')) + config['Architecture']["Head"]['out_channels'] = char_num + model = build_model(config['Architecture']) + + flops = paddle.flops(model, [1, 3, 640, 640]) + logger.info(f"FLOPs before pruning: {flops}") + + from paddleslim.dygraph import FPGMFilterPruner + model.train() + pruner = FPGMFilterPruner(model, [1, 3, 640, 640]) + + # build loss + loss_class = build_loss(config['Loss']) + + # build optim + optimizer, lr_scheduler = build_optimizer( + config['Optimizer'], + epochs=config['Global']['epoch_num'], + step_each_epoch=len(train_dataloader), + parameters=model.parameters()) + + # build metric + eval_class = build_metric(config['Metric']) + # load pretrain model + pre_best_model_dict = init_model(config, model, logger, optimizer) + + logger.info('train dataloader has {} iters, valid dataloader has {} iters'. + format(len(train_dataloader), len(valid_dataloader))) + # build metric + eval_class = build_metric(config['Metric']) + + logger.info('train dataloader has {} iters, valid dataloader has {} iters'. + format(len(train_dataloader), len(valid_dataloader))) + + def eval_fn(): + metric = program.eval(model, valid_dataloader, post_process_class, + eval_class) + logger.info(f"metric['hmean']: {metric['hmean']}") + return metric['hmean'] + + params_sensitive = pruner.sensitive( + eval_func=eval_fn, + sen_file="./sen.pickle", + skip_vars=[ + "conv2d_57.w_0", "conv2d_transpose_2.w_0", "conv2d_transpose_3.w_0" + ]) + + logger.info( + "The sensitivity analysis results of model parameters saved in sen.pickle" + ) + # calculate pruned params's ratio + params_sensitive = pruner._get_ratios_by_loss(params_sensitive, loss=0.02) + for key in params_sensitive.keys(): + logger.info(f"{key}, {params_sensitive[key]}") + + plan = pruner.prune_vars(params_sensitive, [0]) + for param in model.parameters(): + if ("weights" in param.name and "conv" in param.name) or ( + "w_0" in param.name and "conv2d" in param.name): + logger.info(f"{param.name}: {param.shape}") + + flops = paddle.flops(model, [1, 3, 640, 640]) + logger.info(f"FLOPs after pruning: {flops}") + + # start train + + program.train(config, train_dataloader, valid_dataloader, device, model, + loss_class, optimizer, lr_scheduler, post_process_class, + eval_class, pre_best_model_dict, logger, vdl_writer) + + +if __name__ == '__main__': + config, device, logger, vdl_writer = program.preprocess(is_train=True) + main(config, device, logger, vdl_writer) diff --git a/doc/doc_ch/FAQ.md b/doc/doc_ch/FAQ.md index 11a9f35d9e8a92c9bef0da68e085739c1b1e821c..2af9006ad19c4f89d4798b581c0ebfedef7984f7 100755 --- a/doc/doc_ch/FAQ.md +++ b/doc/doc_ch/FAQ.md @@ -9,43 +9,38 @@ ## PaddleOCR常见问题汇总(持续更新) -* [近期更新（2021.1.25）](#近期更新) +* [近期更新（2021.2.1）](#近期更新) * [【精选】OCR精选10个问题](#OCR精选10个问题) * [【理论篇】OCR通用32个问题](#OCR通用问题) * [基础知识7题](#基础知识) * [数据集7题](#数据集2) * [模型训练调优18题](#模型训练调优2) -* [【实战篇】PaddleOCR实战115个问题](#PaddleOCR实战问题) +* [【实战篇】PaddleOCR实战120个问题](#PaddleOCR实战问题) * [使用咨询38题](#使用咨询) - * [数据集17题](#数据集3) - * [模型训练调优28题](#模型训练调优3) - * [预测部署32题](#预测部署3) + * [数据集18题](#数据集3) + * [模型训练调优30题](#模型训练调优3) + * [预测部署34题](#预测部署3) -## 近期更新（2021.1.25） +## 近期更新（2021.2.1） -#### Q3.1.37: 小语种模型只有识别模型，没有检测模型吗？ +#### Q3.2.18: PaddleOCR动态图版本如何finetune？ +**A**：finetune需要将配置文件里的 Global.load_static_weights设置为false，如果没有此字段可以手动添加，然后将模型地址放到Global.pretrained_model字段下即可。 -**A**：小语种（包括纯英文数字）的检测模型和中文的检测模型是共用的，在训练中文检测模型时加入了多语言数据。https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_en/models_list_en.md#1-text-detection-model。 -#### Q3.1.38: module 'paddle.distributed' has no attribute ‘get_rank’。 +#### Q3.3.29: 微调v1.1预训练的模型，可以直接用文字垂直排列和上下颠倒的图片吗？还是必须要水平排列的？ +**A**：1.1和2.0的模型一样，微调时，垂直排列的文字需要逆时针旋转 90° 后加入训练，上下颠倒的需要旋转为水平的。 -**A**：Paddle版本问题，请安装2.0版本Paddle：pip install paddlepaddle==2.0.0。 +#### Q3.3.30: 模型训练过程中如何得到 best_accuracy 模型？ +**A**：配置文件里的eval_batch_step字段用来控制多少次iter进行一次eval，在eval完成后会自动生成 best_accuracy 模型，所以如果希望很快就能拿到best_accuracy模型，可以将eval_batch_step改小一点(例如，10)。 -#### Q3.4.30: PaddleOCR是否支持在华为鲲鹏920CPU上部署？ +#### Q3.4.33: 如何多进程运行paddleocr？ +**A**：实例化多个paddleocr服务，然后将服务注册到注册中心，之后通过注册中心统一调度即可，关于注册中心，可以搜索eureka了解一下具体使用，其他的注册中心也行。 -**A**：目前Paddle的预测库是支持华为鲲鹏920CPU的，但是OCR还没在这些芯片上测试过，可以自己调试，有问题反馈给我们。 -#### Q3.4.31: 采用Paddle-Lite进行端侧部署，出现问题，环境没问题。 - -**A**：如果你的预测库是自己编译的，那么你的nb文件也要自己编译，用同一个lite版本。不能直接用下载的nb文件，因为版本不同。 - -#### Q3.4.32: PaddleOCR的模型支持onnx转换吗？ - -**A**：我们目前已经通过Paddle2ONNX来支持各模型套件的转换，PaddleOCR基于PaddlePaddle 2.0的版本（dygraph分支）已经支持导出为ONNX，欢迎关注Paddle2ONNX，了解更多项目的进展： -Paddle2ONNX项目：https://github.com/PaddlePaddle/Paddle2ONNX -Paddle2ONNX支持转换的[模型列表](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/docs/zh/model_zoo.md#%E5%9B%BE%E5%83%8Focr) +#### Q3.4.34: 2.0训练出来的模型，能否在1.1版本上进行部署？ +**A**：这个是不建议的，2.0训练出来的模型建议使用dygraph分支里提供的部署代码。 ## 【精选】OCR精选10个问题 @@ -366,13 +361,13 @@ Paddle2ONNX支持转换的[模型列表](https://github.com/PaddlePaddle/Paddle2 （2）inference模型下载时，如果没有安装wget，可直接点击模型链接或将链接地址复制到浏览器进行下载，并解压放置到相应目录。 #### Q3.1.17：PaddleOCR开源的超轻量模型和通用OCR模型的区别？ -**A**：目前PaddleOCR开源了2个中文模型，分别是8.6M超轻量中文模型和通用中文OCR模型。两者对比信息如下： +**A**：目前PaddleOCR开源了2个中文模型，分别是9.4M超轻量中文模型和通用中文OCR模型。两者对比信息如下： - 相同点：两者使用相同的**算法**和**训练数据**； - 不同点：不同之处在于**骨干网络**和**通道参数**，超轻量模型使用MobileNetV3作为骨干网络，通用模型使用Resnet50_vd作为检测模型backbone，Resnet34_vd作为识别模型backbone，具体参数差异可对比两种模型训练的配置文件. |模型|骨干网络|检测训练配置|识别训练配置| |-|-|-|-| -|8.6M超轻量中文OCR模型|MobileNetV3+MobileNetV3|det_mv3_db.yml|rec_chinese_lite_train.yml| +|9.4M超轻量中文OCR模型|MobileNetV3+MobileNetV3|det_mv3_db.yml|rec_chinese_lite_train.yml| |通用中文OCR模型|Resnet50_vd+Resnet34_vd|det_r50_vd_db.yml|rec_chinese_common_train.yml| #### Q3.1.18：如何加入自己的检测算法？ @@ -578,6 +573,9 @@ StyleText的用途主要是：提取style_image中的字体、背景等style信 **A**：PPOCRLabel可运行于Linux、Windows、MacOS等多种系统。操作步骤可以参考文档，https://github.com/PaddlePaddle/PaddleOCR/blob/develop/PPOCRLabel/README.md +#### Q3.2.18: PaddleOCR动态图版本如何finetune？ +**A**：finetune需要将配置文件里的 Global.load_static_weights设置为false，如果没有此字段可以手动添加，然后将模型地址放到Global.pretrained_model字段下即可。 + ### 模型训练调优 @@ -723,6 +721,12 @@ ps -axu | grep train.py | awk '{print $2}' | xargs kill -9 **A**：可以参考[配置文件](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)在Train['dataset']['transforms']添加RecAug字段，使数据增强生效。可以通过添加对aug_prob设置，表示每种数据增强采用的概率。aug_prob默认是0.4.由于tia数据增强特殊性，默认不采用，可以通过添加use_tia设置，使tia数据增强生效。详细设置可以参考[ISSUE 1744](https://github.com/PaddlePaddle/PaddleOCR/issues/1744)。 +#### Q3.3.29: 微调v1.1预训练的模型，可以直接用文字垂直排列和上下颠倒的图片吗？还是必须要水平排列的？ +**A**：1.1和2.0的模型一样，微调时，垂直排列的文字需要逆时针旋转 90°后加入训练，上下颠倒的需要旋转为水平的。 + +#### Q3.3.30: 模型训练过程中如何得到 best_accuracy 模型？ +**A**：配置文件里的eval_batch_step字段用来控制多少次iter进行一次eval，在eval完成后会自动生成 best_accuracy 模型，所以如果希望很快就能拿到best_accuracy模型，可以将eval_batch_step改小一点，如改为[10,10]，这样表示第10次迭代后，以后没隔10个迭代就进行一次模型的评估。 + ### 预测部署 @@ -878,3 +882,10 @@ img = cv.imdecode(img_array, -1) **A**：我们目前已经通过Paddle2ONNX来支持各模型套件的转换，PaddleOCR基于PaddlePaddle 2.0的版本（dygraph分支）已经支持导出为ONNX，欢迎关注Paddle2ONNX，了解更多项目的进展： Paddle2ONNX项目：https://github.com/PaddlePaddle/Paddle2ONNX Paddle2ONNX支持转换的[模型列表](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/docs/zh/model_zoo.md#%E5%9B%BE%E5%83%8Focr) + + +#### Q3.4.33: 如何多进程运行paddleocr？ +**A**：实例化多个paddleocr服务，然后将服务注册到注册中心，之后通过注册中心统一调度即可，关于注册中心，可以搜索eureka了解一下具体使用，其他的注册中心也行。 + +#### Q3.4.34: 2.0训练出来的模型，能否在1.1版本上进行部署？ +**A**：这个是不建议的，2.0训练出来的模型建议使用dygraph分支里提供的部署代码。 diff --git a/doc/doc_ch/algorithm_overview.md b/doc/doc_ch/algorithm_overview.md index 59d1bc8c444e3a70bbea83f87afcbd2f5cf44191..c8fc280d80056395bbc841a973004b06844b1214 100755 --- a/doc/doc_ch/algorithm_overview.md +++ b/doc/doc_ch/algorithm_overview.md @@ -40,8 +40,8 @@ PaddleOCR基于动态图开源的文本识别算法列表： - [x] CRNN([paper](https://arxiv.org/abs/1507.05717))[7]（ppocr推荐） - [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))[10] - [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11] -- [ ] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] coming soon -- [ ] SRN([paper](https://arxiv.org/abs/2003.12294))[5] coming soon +- [x] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] +- [x] SRN([paper](https://arxiv.org/abs/2003.12294))[5] 参考[DTRB][3](https://arxiv.org/abs/1904.01906)文字识别训练和评估流程，使用MJSynth和SynthText两个文字识别数据集训练，在IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE数据集上进行评估，算法效果如下： @@ -53,5 +53,9 @@ PaddleOCR基于动态图开源的文本识别算法列表： |CRNN|MobileNetV3|79.97%|rec_mv3_none_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_bilstm_ctc_v2.0_train.tar)| |StarNet|Resnet34_vd|84.44%|rec_r34_vd_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_ctc_v2.0_train.tar)| |StarNet|MobileNetV3|81.42%|rec_mv3_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_ctc_v2.0_train.tar)| +|RARE|MobileNetV3|82.5%|rec_mv3_tps_bilstm_att |[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_att_v2.0_train.tar)| +|RARE|Resnet34_vd|83.6%|rec_r34_vd_tps_bilstm_att |[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_att_v2.0_train.tar)| +|SRN|Resnet50_vd_fpn| 88.52% | rec_r50fpn_vd_none_srn | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r50_vd_srn_train.tar) | + PaddleOCR文本识别算法的训练和使用请参考文档教程中[模型训练/评估中的文本识别部分](./recognition.md)。 diff --git a/doc/doc_ch/angle_class.md b/doc/doc_ch/angle_class.md index 6e68134a4d1b8d9b8927d67c9724ba88563383a4..ad25a6661817623419af0c0c7a139dd4bfaeb08c 100644 --- a/doc/doc_ch/angle_class.md +++ b/doc/doc_ch/angle_class.md @@ -1,4 +1,12 @@ ## 文字角度分类 +### 方法介绍 +文字角度分类主要用于图片非0度的场景下，在这种场景下需要对图片里检测到的文本行进行一个转正的操作。在PaddleOCR系统内， +文字检测之后得到的文本行图片经过仿射变换之后送入识别模型，此时只需要对文字进行一个0和180度的角度分类，因此PaddleOCR内置的 +文字角度分类器**只支持了0和180度的分类**。如果想支持更多角度，可以自己修改算法进行支持。 + +0和180度数据样本例子： + +![](../imgs_results/angle_class_example.jpg) ### 数据准备 @@ -13,7 +21,7 @@ ln -sf /train_data/cls/dataset 请参考下文组织您的数据。 - 训练集 -首先请将训练图片放入同一个文件夹（train_images），并用一个txt文件（cls_gt_train.txt）记录图片路径和标签。 +首先建议将训练图片放入同一个文件夹，并用一个txt文件（cls_gt_train.txt）记录图片路径和标签。 **注意：** 默认请将图片路径和图片标签用 `\t` 分割，如用其他方式分割将造成训练报错 @@ -21,8 +29,8 @@ ln -sf /train_data/cls/dataset ``` " 图像文件名 图像标注信息 " -train/word_001.jpg 0 -train/word_002.jpg 180 +train/cls/train/word_001.jpg 0 +train/cls/train/word_002.jpg 180 ``` 最终训练集应有如下文件结构： diff --git a/doc/doc_ch/inference.md b/doc/doc_ch/inference.md index c4601e1526d29e0a8c62030a4b47d2b2cc193d5d..7968b355ea936d465b3c173c0fcdb3e08f12f16e 100755 --- a/doc/doc_ch/inference.md +++ b/doc/doc_ch/inference.md @@ -2,10 +2,11 @@ # 基于Python预测引擎推理 inference 模型（`paddle.jit.save`保存的模型） -一般是模型训练完成后保存的固化模型，多用于预测部署。训练过程中保存的模型是checkpoints模型，保存的是模型的参数，多用于恢复训练等。 -与checkpoints模型相比，inference 模型会额外保存模型的结构信息，在预测部署、加速推理上性能优越，灵活方便，适合与实际系统集成。 +一般是模型训练，把模型结构和模型参数保存在文件中的固化模型，多用于预测部署场景。 +训练过程中保存的模型是checkpoints模型，保存的只有模型的参数，多用于恢复训练等。 +与checkpoints模型相比，inference 模型会额外保存模型的结构信息，在预测部署、加速推理上性能优越，灵活方便，适合于实际系统集成。 -接下来首先介绍如何将训练的模型转换成inference模型，然后将依次介绍文本检测、文本角度分类器、文本识别以及三者串联基于预测引擎推理。 +接下来首先介绍如何将训练的模型转换成inference模型，然后将依次介绍文本检测、文本角度分类器、文本识别以及三者串联在CPU、GPU上的预测方法。 - [一、训练模型转inference模型](#训练模型转inference模型) @@ -22,8 +23,9 @@ inference 模型（`paddle.jit.save`保存的模型） - [三、文本识别模型推理](#文本识别模型推理) - [1. 超轻量中文识别模型推理](#超轻量中文识别模型推理) - [2. 基于CTC损失的识别模型推理](#基于CTC损失的识别模型推理) - - [3. 自定义文本识别字典的推理](#自定义文本识别字典的推理) - - [4. 多语言模型的推理](#多语言模型的推理) + - [3. 基于SRN损失的识别模型推理](#基于SRN损失的识别模型推理) + - [4. 自定义文本识别字典的推理](#自定义文本识别字典的推理) + - [5. 多语言模型的推理](#多语言模型的推理) - [四、方向分类模型推理](#方向识别模型推理) - [1. 方向分类模型推理](#方向分类模型推理) @@ -139,7 +141,7 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_m ![](../imgs_results/det_res_00018069.jpg) 通过参数`limit_type`和`det_limit_side_len`来对图片的尺寸进行限制， -`litmit_type`可选参数为[`max`, `min`]， +`limit_type`可选参数为[`max`, `min`]， `det_limit_size_len` 为正整数，一般设置为32 的倍数，比如960。 参数默认设置为`limit_type='max', det_limit_side_len=960`。表示网络输入图像的最长边不能超过960， @@ -295,8 +297,20 @@ Predicts of ./doc/imgs_words_en/word_336.png:('super', 0.9999073) self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz" dict_character = list(self.character_str) ``` + +### 3. 基于SRN损失的识别模型推理 +基于SRN损失的识别模型，需要额外设置识别算法参数 --rec_algorithm="SRN"。 +同时需要保证预测shape与训练时一致，如： --rec_image_shape="1, 64, 256" -### 3. 自定义文本识别字典的推理 +``` +python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" \ + --rec_model_dir="./inference/srn/" \ + --rec_image_shape="1, 64, 256" \ + --rec_char_type="en" \ + --rec_algorithm="SRN" +``` + +### 4. 自定义文本识别字典的推理 如果训练时修改了文本的字典，在使用inference模型预测时，需要通过`--rec_char_dict_path`指定使用的字典路径，并且设置 `rec_char_type=ch` ``` @@ -304,7 +318,7 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png ``` -### 4. 多语言模型的推理 +### 5. 多语言模型的推理 如果您需要预测的是其他语言模型，在使用inference模型预测时，需要通过`--rec_char_dict_path`指定使用的字典路径, 同时为了得到正确的可视化结果， 需要通过 `--vis_font_path` 指定可视化的字体路径，`doc/fonts/` 路径下有默认提供的小语种字体，例如韩文识别： diff --git a/doc/doc_ch/installation.md b/doc/doc_ch/installation.md index fce151eb9fee567477c09eee211633f7377dddb3..7e7523b999aa6fdee9bbaa4fb388655e01c28c16 100644 --- a/doc/doc_ch/installation.md +++ b/doc/doc_ch/installation.md @@ -30,7 +30,7 @@ sudo nvidia-docker run --name ppocr -v $PWD:/paddle --shm-size=64G --network=hos sudo docker container exec -it ppocr /bin/bash ``` -**2. 安装PaddlePaddle Fluid v2.0** +**2. 安装PaddlePaddle 2.0** ``` pip3 install --upgrade pip diff --git a/doc/doc_ch/models_list.md b/doc/doc_ch/models_list.md index fbfb3838b7f860678b10ef4507ebf6c0d4b815c9..d647d032674693466525d12c25b15c961e0eb86f 100644 --- a/doc/doc_ch/models_list.md +++ b/doc/doc_ch/models_list.md @@ -1,5 +1,9 @@ ## OCR模型列表（V2.0，2021年1月20日更新） -**说明** ：2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md)的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。 + +> **说明** +> 1. 2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md)的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。 +> 2. 本文档提供的是PPOCR自研模型列表，更多基于公开数据集的算法介绍与预训练模型可以参考：[算法概览文档](./algorithm_overview.md)。 + - [一、文本检测模型](#文本检测模型) - [二、文本识别模型](#文本识别模型) @@ -12,9 +16,14 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训 |模型类型|模型格式|简介| |--- | --- | --- | -|推理模型|inference.pdmodel、inference.pdiparams|用于python预测引擎推理，[详情](./inference.md)| +|推理模型|inference.pdmodel、inference.pdiparams|用于预测引擎推理，[详情](./inference.md)| |训练模型、预训练模型|\*.pdparams、\*.pdopt、\*.states |训练过程中保存的模型的参数、优化器状态和训练中间信息，多用于模型指标评估和恢复训练| -|slim模型|\*.nb|用于lite部署| +|slim模型|\*.nb|经过飞桨模型压缩工具PaddleSlim压缩后的模型，适用于移动端/IoT端等端侧部署场景（需使用飞桨Paddle Lite部署）。| + + +各个模型的关系如下面的示意图所示。 + +![](../imgs/model_prod_flow_ch.png) diff --git a/doc/doc_ch/recognition.md b/doc/doc_ch/recognition.md index c5f459bdb88558b1cdea93b9b85eed0e4bb8433b..907cf24e1a31104096ab6c0cf0819457852d1490 100644 --- a/doc/doc_ch/recognition.md +++ b/doc/doc_ch/recognition.md @@ -1,60 +1,94 @@ ## 文字识别 -- [一、数据准备](#数据准备) - - [数据下载](#数据下载) - - [自定义数据集](#自定义数据集) - - [字典](#字典) - - [支持空格](#支持空格) +- [1 数据准备](#数据准备) + - [1.1 自定义数据集](#自定义数据集) + - [1.2 数据下载](#数据下载) + - [1.3 字典](#字典) + - [1.4 支持空格](#支持空格) -- [二、启动训练](#启动训练) - - [1. 数据增强](#数据增强) - - [2. 训练](#训练) - - [3. 小语种](#小语种) +- [2 启动训练](#启动训练) + - [2.1 数据增强](#数据增强) + - [2.2 训练](#训练) + - [2.3 小语种](#小语种) -- [三、评估](#评估) +- [3 评估](#评估) -- [四、预测](#预测) - - [1. 训练引擎预测](#训练引擎预测) +- [4 预测](#预测) + - [4.1 训练引擎预测](#训练引擎预测) -### 数据准备 +### 1. 数据准备 -PaddleOCR 支持两种数据格式: `lmdb` 用于训练公开数据，调试算法; `通用数据` 训练自己的数据: - -请按如下步骤设置数据集： +PaddleOCR 支持两种数据格式: + - `lmdb` 用于训练以lmdb格式存储的数据集; + - `通用数据` 用于训练以文本文件存储的数据集: 训练数据的默认存储路径是 `PaddleOCR/train_data`,如果您的磁盘上已有数据集，只需创建软链接至数据集目录： ``` +# linux and mac os ln -sf /train_data/dataset +# windows +mklink /d /train_data/dataset ``` - -* 数据下载 - -若您本地没有数据集，可以在官网下载 [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here)，下载 benchmark 所需的lmdb格式数据集。 + +#### 1.1 自定义数据集 +下面以通用数据集为例， 介绍如何准备数据集： - -* 使用自己数据集 +* 训练集 -若您希望使用自己的数据进行训练，请参考下文组织您的数据。 +建议将训练图片放入同一个文件夹，并用一个txt文件（rec_gt_train.txt）记录图片路径和标签，txt文件里的内容如下: -- 训练集 +**注意：** txt文件中默认请将图片路径和图片标签用 \t 分割，如用其他方式分割将造成训练报错。 -首先请将训练图片放入同一个文件夹（train_images），并用一个txt文件（rec_gt_train.txt）记录图片路径和标签。 +``` +" 图像文件名 图像标注信息 " -**注意：** 默认请将图片路径和图片标签用 \t 分割，如用其他方式分割将造成训练报错 +train_data/rec/train/word_001.jpg 简单可依赖 +train_data/rec/train/word_002.jpg 用科技让复杂的世界更简单 +... +``` +最终训练集应有如下文件结构： +``` +|-train_data + |-rec + |- rec_gt_train.txt + |- train + |- word_001.png + |- word_002.jpg + |- word_003.jpg + | ... ``` -" 图像文件名 图像标注信息 " -train_data/train_0001.jpg 简单可依赖 -train_data/train_0002.jpg 用科技让复杂的世界更简单 +- 测试集 + +同训练集类似，测试集也需要提供一个包含所有图片的文件夹（test）和一个rec_gt_test.txt，测试集的结构如下所示： + ``` -PaddleOCR 提供了一份用于训练 icdar2015 数据集的标签文件，通过以下方式下载： +|-train_data + |-rec + |- rec_gt_test.txt + |- test + |- word_001.jpg + |- word_002.jpg + |- word_003.jpg + | ... +``` + + + +1.2 数据下载 + +若您本地没有数据集，可以在官网下载 [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，下载 benchmark 所需的lmdb格式数据集。 + +如果你使用的是icdar2015的公开数据集，PaddleOCR 提供了一份用于训练 icdar2015 数据集的标签文件，通过以下方式下载： + +如果希望复现SRN的论文指标，需要下载离线[增广数据](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA),提取码: y3ry。增广数据是由MJSynth和SynthText做旋转和扰动得到的。数据下载完成后请解压到 {your_path}/PaddleOCR/train_data/data_lmdb_release/training/ 路径下。 ``` # 训练集标签 @@ -70,34 +104,8 @@ PaddleOCR 也提供了数据格式转换脚本，可以将官网 label 转换支 python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_label="rec_gt_label.txt" ``` -最终训练集应有如下文件结构： -``` -|-train_data - |-ic15_data - |- rec_gt_train.txt - |- train - |- word_001.png - |- word_002.jpg - |- word_003.jpg - | ... -``` - -- 测试集 - -同训练集类似，测试集也需要提供一个包含所有图片的文件夹（test）和一个rec_gt_test.txt，测试集的结构如下所示： - -``` -|-train_data - |-ic15_data - |- rec_gt_test.txt - |- test - |- word_001.jpg - |- word_002.jpg - |- word_003.jpg - | ... -``` -- 字典 +1.3 字典 最后需要提供一个字典（{word_dict_name}.txt），使模型在训练时，可以将所有出现的字符映射为字典的索引。 @@ -114,6 +122,10 @@ n word_dict.txt 每行有一个单字，将字符与数字索引映射在一起，“and” 将被映射成 [2 5 1] +* 内置字典 + +PaddleOCR内置了一部分字典，可以按需使用。 + `ppocr/utils/ppocr_keys_v1.txt` 是一个包含6623个字符的中文字典 `ppocr/utils/ic15_dict.txt` 是一个包含36个字符的英文字典 @@ -129,10 +141,10 @@ word_dict.txt 每行有一个单字，将字符与数字索引映射在一起， `ppocr/utils/dict/en_dict.txt` 是一个包含63个字符的英文字典 -您可以按需使用。 + 目前的多语言模型仍处在demo阶段，会持续优化模型并补充语种，**非常欢迎您为我们提供其他语言的字典和字体**， -如您愿意可将字典文件提交至 [dict](../../ppocr/utils/dict) 将语料文件提交至[corpus](../../ppocr/utils/corpus)，我们会在Repo中感谢您。 +如您愿意可将字典文件提交至 [dict](../../ppocr/utils/dict)，我们会在Repo中感谢您。 - 自定义字典 @@ -140,13 +152,13 @@ word_dict.txt 每行有一个单字，将字符与数字索引映射在一起， 并将 `character_type` 设置为 `ch`。 -- 添加空格类别 +1.4 添加空格类别 如果希望支持识别"空格"类别, 请将yml文件中的 `use_space_char` 字段设置为 `True`。 -### 启动训练 +### 2. 启动训练 PaddleOCR提供了训练脚本、评估脚本和预测脚本，本节将以 CRNN 识别模型为例： @@ -171,7 +183,7 @@ tar -xf rec_mv3_none_bilstm_ctc_v2.0_train.tar && rm -rf rec_mv3_none_bilstm_ctc python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml ``` -- 数据增强 +#### 2.1 数据增强 PaddleOCR提供了多种数据增强方式，如果您希望在训练时加入扰动，请在配置文件中设置 `distort: true`。 @@ -182,7 +194,7 @@ PaddleOCR提供了多种数据增强方式，如果您希望在训练时加入 *由于OpenCV的兼容性问题，扰动操作暂时只支持Linux* -- 训练 +#### 2.2 训练 PaddleOCR支持训练和评估交替进行, 可以在 `configs/rec/rec_icdar15_train.yml` 中修改 `eval_batch_step` 设置评估频率，默认每500个iter评估一次。评估过程中默认将最佳acc模型，保存为 `output/rec_CRNN/best_accuracy` 。 @@ -200,6 +212,9 @@ PaddleOCR支持训练和评估交替进行, 可以在 `configs/rec/rec_icdar15_t | rec_mv3_none_none_ctc.yml | Rosetta | Mobilenet_v3 large 0.5 | None | None | ctc | | rec_r34_vd_none_bilstm_ctc.yml | CRNN | Resnet34_vd | None | BiLSTM | ctc | | rec_r34_vd_none_none_ctc.yml | Rosetta | Resnet34_vd | None | None | ctc | +| rec_mv3_tps_bilstm_att.yml | CRNN | Mobilenet_v3 | TPS | BiLSTM | att | +| rec_r34_vd_tps_bilstm_att.yml | CRNN | Resnet34_vd | TPS | BiLSTM | att | +| rec_r50fpn_vd_none_srn.yml | SRN | Resnet50_fpn_vd | None | rnn | srn | 训练中文数据，推荐使用[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)，如您希望尝试其他算法在中文数据集上的效果，请参考下列说明修改配置文件： @@ -268,7 +283,7 @@ Eval: **注意，预测/评估时的配置文件请务必与训练一致。** -- 小语种 +#### 2.3 小语种 PaddleOCR目前已支持26种（除中文外）语种识别，`configs/rec/multi_languages` 路径下提供了一个多语言的配置文件模版: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。 @@ -411,7 +426,7 @@ Eval: ... ``` -### 评估 +### 3 评估 评估数据集可以通过 `configs/rec/rec_icdar15_train.yml` 修改Eval中的 `label_file_path` 设置。 @@ -421,10 +436,10 @@ python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec ``` -### 预测 +### 4 预测 -* 训练引擎的预测 +#### 4.1 训练引擎的预测 使用 PaddleOCR 训练好的模型，可以通过以下脚本进行快速预测。 diff --git a/doc/doc_ch/whl.md b/doc/doc_ch/whl.md index 6b218e31ecdc3d07132a2a88ad22528ed6ef23b4..032d7ae642ad7b2a10253dbfd6310bd5299fb5a0 100644 --- a/doc/doc_ch/whl.md +++ b/doc/doc_ch/whl.md @@ -1,8 +1,8 @@ # paddleocr package使用说明 -## 快速上手 +## 1 快速上手 -### 安装whl包 +### 1.1 安装whl包 pip安装 ```bash @@ -14,9 +14,12 @@ pip install "paddleocr>=2.0.1" # 推荐使用2.0.1+版本 python3 setup.py bdist_wheel pip3 install dist/paddleocr-x.x.x-py3-none-any.whl # x.x.x是paddleocr的版本号 ``` -### 1. 代码使用 -* 检测+分类+识别全流程 +## 2 使用 +### 2.1 代码使用 +paddleocr whl包会自动下载ppocr轻量级模型作为默认模型，可以根据第3节**自定义模型**进行自定义更换。 + +* 检测+方向分类器+识别全流程 ```python from paddleocr import PaddleOCR, draw_ocr # Paddleocr目前支持中英文、英文、法语、德语、韩语、日语，可以通过修改lang参数进行切换 @@ -84,7 +87,7 @@ im_show.save('result.jpg') -* 分类+识别 +* 方向分类器+识别 ```python from paddleocr import PaddleOCR ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory @@ -143,7 +146,7 @@ for line in result: ['韩国小馆', 0.9907421] ``` -* 单独执行分类 +* 单独执行方向分类器 ```python from paddleocr import PaddleOCR ocr = PaddleOCR(use_angle_cls=True) # need to run only once to download and load model into memory @@ -157,14 +160,14 @@ for line in result: ['0', 0.9999924] ``` -### 通过命令行使用 +### 2.2 通过命令行使用 查看帮助信息 ```bash paddleocr -h ``` -* 检测+分类+识别全流程 +* 检测+方向分类器+识别全流程 ```bash paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --use_angle_cls true ``` @@ -188,7 +191,7 @@ paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg ...... ``` -* 分类+识别 +* 方向分类器+识别 ```bash paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false ``` @@ -220,7 +223,7 @@ paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --det false ['韩国小馆', 0.9907421] ``` -* 单独执行分类 +* 单独执行方向分类器 ```bash paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls true --det false --rec false ``` @@ -230,11 +233,11 @@ paddleocr --image_dir PaddleOCR/doc/imgs_words/ch/word_1.jpg --use_angle_cls tru ['0', 0.9999924] ``` -## 自定义模型 +## 3 自定义模型 当内置模型无法满足需求时，需要使用到自己训练的模型。 首先，参照[inference.md](./inference.md) 第一节转换将检测、分类和识别模型转换为inference模型，然后按照如下方式使用 -### 代码使用 +### 3.1 代码使用 ```python from paddleocr import PaddleOCR, draw_ocr # 模型路径下必须含有model和params文件 @@ -255,17 +258,17 @@ im_show = Image.fromarray(im_show) im_show.save('result.jpg') ``` -### 通过命令行使用 +### 3.2 通过命令行使用 ```bash paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --det_model_dir {your_det_model_dir} --rec_model_dir {your_rec_model_dir} --rec_char_dict_path {your_rec_char_dict_path} --cls_model_dir {your_cls_model_dir} --use_angle_cls true ``` -### 使用网络图片或者numpy数组作为输入 +## 4 使用网络图片或者numpy数组作为输入 -1. 网络图片 +### 4.1 网络图片 -代码使用 +- 代码使用 ```python from paddleocr import PaddleOCR, draw_ocr # Paddleocr目前支持中英文、英文、法语、德语、韩语、日语，可以通过修改lang参数进行切换 @@ -286,12 +289,12 @@ im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc im_show = Image.fromarray(im_show) im_show.save('result.jpg') ``` -命令行模式 +- 命令行模式 ```bash paddleocr --image_dir http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg --use_angle_cls=true ``` -2. numpy数组 +### 4.2 numpy数组 仅通过代码使用时支持numpy数组作为输入 ```python from paddleocr import PaddleOCR, draw_ocr @@ -301,7 +304,7 @@ ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to downlo img_path = 'PaddleOCR/doc/imgs/11.jpg' img = cv2.imread(img_path) # img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY), 如果你自己训练的模型支持灰度图，可以将这句话的注释取消 -result = ocr.ocr(img_path, cls=True) +result = ocr.ocr(img, cls=True) for line in result: print(line) @@ -316,7 +319,7 @@ im_show = Image.fromarray(im_show) im_show.save('result.jpg') ``` -## 参数说明 +## 5 参数说明 | 字段 | 说明 | 默认值 | |-------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------| diff --git a/doc/doc_en/algorithm_overview_en.md b/doc/doc_en/algorithm_overview_en.md index 68bfd529972183208220b1c87227639d683fea62..77b9642e3b880547b1df6620d931689982db6d29 100755 --- a/doc/doc_en/algorithm_overview_en.md +++ b/doc/doc_en/algorithm_overview_en.md @@ -42,8 +42,8 @@ PaddleOCR open-source text recognition algorithms list: - [x] CRNN([paper](https://arxiv.org/abs/1507.05717))[7] - [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))[10] - [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11] -- [ ] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] coming soon -- [ ] SRN([paper](https://arxiv.org/abs/2003.12294))[5] coming soon +- [x] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] +- [x] SRN([paper](https://arxiv.org/abs/2003.12294))[5] Refer to [DTRB](https://arxiv.org/abs/1904.01906), the training and evaluation result of these above text recognition (using MJSynth and SynthText for training, evaluate on IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE) is as follow: @@ -55,5 +55,8 @@ Refer to [DTRB](https://arxiv.org/abs/1904.01906), the training and evaluation r |CRNN|MobileNetV3|79.97%|rec_mv3_none_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_bilstm_ctc_v2.0_train.tar)| |StarNet|Resnet34_vd|84.44%|rec_r34_vd_tps_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_ctc_v2.0_train.tar)| |StarNet|MobileNetV3|81.42%|rec_mv3_tps_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_ctc_v2.0_train.tar)| +|RARE|MobileNetV3|82.5%|rec_mv3_tps_bilstm_att |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_att_v2.0_train.tar)| +|RARE|Resnet34_vd|83.6%|rec_r34_vd_tps_bilstm_att |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_att_v2.0_train.tar)| +|SRN|Resnet50_vd_fpn| 88.52% | rec_r50fpn_vd_none_srn |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r50_vd_srn_train.tar)| Please refer to the document for training guide and use of PaddleOCR text recognition algorithms [Text recognition model training/evaluation/prediction](./recognition_en.md) diff --git a/doc/doc_en/angle_class_en.md b/doc/doc_en/angle_class_en.md index d1cc712f312bf8e70c0b399422519217df323129..0044d85ac0a43529c67746d25118bd80ee52be9a 100644 --- a/doc/doc_en/angle_class_en.md +++ b/doc/doc_en/angle_class_en.md @@ -1,5 +1,12 @@ ## TEXT ANGLE CLASSIFICATION +### Method introduction +The angle classification is used in the scene where the image is not 0 degrees. In this scene, it is necessary to perform a correction operation on the text line detected in the picture. In the PaddleOCR system, +The text line image obtained after text detection is sent to the recognition model after affine transformation. At this time, only a 0 and 180 degree angle classification of the text is required, so the built-in PaddleOCR text angle classifier **only supports 0 and 180 degree classification**. If you want to support more angles, you can modify the algorithm yourself to support. + +Example of 0 and 180 degree data samples： + +![](../imgs_results/angle_class_example.jpg) ### DATA PREPARATION Please organize the dataset as follows: diff --git a/doc/doc_en/inference_en.md b/doc/doc_en/inference_en.md index ccbb71847d5946e854b88817a162957af0e6ed00..aa3e0536cb6a4cb73f0388293ce72183c62e87a1 100755 --- a/doc/doc_en/inference_en.md +++ b/doc/doc_en/inference_en.md @@ -5,7 +5,8 @@ The inference model (the model saved by `paddle.jit.save`) is generally a solidi The model saved during the training process is the checkpoints model, which saves the parameters of the model and is mostly used to resume training. -Compared with the checkpoints model, the inference model will additionally save the structural information of the model. It has superior performance in predicting in deployment and accelerating inferencing, is flexible and convenient, and is suitable for integration with actual systems. For more details, please refer to the document [Classification Framework](https://github.com/PaddlePaddle/PaddleClas/blob/master/docs/zh_CN/extension/paddle_inference.md). +Compared with the checkpoints model, the inference model will additionally save the structural information of the model. Therefore, it is easier to deploy because the model structure and model parameters are already solidified in the inference model file, and is suitable for integration with actual systems. +For more details, please refer to the document [Classification Framework](https://github.com/PaddlePaddle/PaddleClas/blob/release%2F2.0/docs/zh_CN/extension/paddle_mobile_inference.md). Next, we first introduce how to convert a trained model into an inference model, and then we will introduce text detection, text recognition, angle class, and the concatenation of them based on inference model. @@ -25,6 +26,7 @@ Next, we first introduce how to convert a trained model into an inference model, - [TEXT RECOGNITION MODEL INFERENCE](#RECOGNITION_MODEL_INFERENCE) - [1. LIGHTWEIGHT CHINESE MODEL](#LIGHTWEIGHT_RECOGNITION) - [2. CTC-BASED TEXT RECOGNITION MODEL INFERENCE](#CTC-BASED_RECOGNITION) + - [3. SRN-BASED TEXT RECOGNITION MODEL INFERENCE](#SRN-BASED_RECOGNITION) - [3. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY](#USING_CUSTOM_CHARACTERS) - [4. MULTILINGUAL MODEL INFERENCE](MULTILINGUAL_MODEL_INFERENCE) @@ -146,7 +148,7 @@ The visual text detection results are saved to the ./inference_results folder by ![](../imgs_results/det_res_00018069.jpg) You can use the parameters `limit_type` and `det_limit_side_len` to limit the size of the input image, -The optional parameters of `litmit_type` are [`max`, `min`], and +The optional parameters of `limit_type` are [`max`, `min`], and `det_limit_size_len` is a positive integer, generally set to a multiple of 32, such as 960. The default setting of the parameters is `limit_type='max', det_limit_side_len=960`. Indicates that the longest side of the network input image cannot exceed 960, @@ -304,8 +306,23 @@ self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz" dict_character = list(self.character_str) ``` + +### 3. SRN-BASED TEXT RECOGNITION MODEL INFERENCE + +The recognition model based on SRN requires additional setting of the recognition algorithm parameter +--rec_algorithm="SRN". At the same time, it is necessary to ensure that the predicted shape is consistent +with the training, such as: --rec_image_shape="1, 64, 256" + +``` +python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" \ + --rec_model_dir="./inference/srn/" \ + --rec_image_shape="1, 64, 256" \ + --rec_char_type="en" \ + --rec_algorithm="SRN" +``` + -### 3. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY +### 4. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY If the text dictionary is modified during training, when using the inference model to predict, you need to specify the dictionary path used by `--rec_char_dict_path`, and set `rec_char_type=ch` ``` @@ -313,7 +330,7 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png ``` -### 4. MULTILINGAUL MODEL INFERENCE +### 5. MULTILINGAUL MODEL INFERENCE If you need to predict other language models, when using inference model prediction, you need to specify the dictionary path used by `--rec_char_dict_path`. At the same time, in order to get the correct visualization results, You need to specify the visual font path through `--vis_font_path`. There are small language fonts provided by default under the `doc/fonts` path, such as Korean recognition: diff --git a/doc/doc_en/installation_en.md b/doc/doc_en/installation_en.md index 35c1881d12087e6509a68b504729d9ef20240e9c..dec384b2f27f8bb36ee67d8b040b532b30e0b028 100644 --- a/doc/doc_en/installation_en.md +++ b/doc/doc_en/installation_en.md @@ -33,7 +33,7 @@ You can also visit [DockerHub](https://hub.docker.com/r/paddlepaddle/paddle/tags sudo docker container exec -it ppocr /bin/bash ``` -**2. Install PaddlePaddle Fluid v2.0** +**2. Install PaddlePaddle 2.0** ``` pip3 install --upgrade pip diff --git a/doc/doc_en/models_list_en.md b/doc/doc_en/models_list_en.md index 33033f8348fa4fb08d6e8998ff53cd62349c214e..e0163972b94ddfde8215cd323e2ba037752d3b4c 100644 --- a/doc/doc_en/models_list_en.md +++ b/doc/doc_en/models_list_en.md @@ -1,5 +1,7 @@ ## OCR model list（V2.0, updated on 2021.1.20） -**Note** : Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance. +> **Note** +> 1. Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance. +> 2. All models in this tutorial are all ppocr-series models, for more introduction of algorithms and models based on public dataset, you can refer to [algorithm overview tutorial](./algorithm_overview_en.md). - [1. Text Detection Model](#Detection) - [2. Text Recognition Model](#Recognition) @@ -12,9 +14,13 @@ The downloadable models provided by PaddleOCR include `inference model`, `traine |model type|model format|description| |--- | --- | --- | -|inference model|inference.pdmodel、inference.pdiparams|Used for reasoning based on Python prediction engine，[detail](./inference_en.md)| +|inference model|inference.pdmodel、inference.pdiparams|Used for inference based on Paddle inference engine，[detail](./inference_en.md)| |trained model, pre-trained model|\*.pdparams、\*.pdopt、\*.states |The checkpoints model saved in the training process, which stores the parameters of the model, mostly used for model evaluation and continuous training.| -|slim model|\*.nb|Generally used for Lite deployment| +|slim model|\*.nb| Model compressed by PaddleSim (a model compression tool using PaddlePaddle), which is suitable for mobile-side deployment scenarios (Paddle-Lite is needed for slim model deployment). | + +Relationship of the above models is as follows. + +![](../imgs_en/model_prod_flow_en.png) ### 1. Text Detection Model @@ -80,7 +86,7 @@ If you want to train your own model, you can prepare the training set file, veri cd {your/path/}PaddleOCR/configs/rec/multi_language/ # The -l or --language parameter is required # --train modify train_list path -# --val modify eval_list path +# --val modify eval_list path # --data_dir modify data dir # -o modify default parameters # --dict Change the dictionary path. The example uses the default dictionary path, so that this parameter can be empty. diff --git a/doc/doc_en/recognition_en.md b/doc/doc_en/recognition_en.md index 22f89cdef080afe0b119d08d1e88f02ede5932c1..aeb9aa0d43e400c4d6e733b2c9f4a74559dccecb 100644 --- a/doc/doc_en/recognition_en.md +++ b/doc/doc_en/recognition_en.md @@ -1,79 +1,70 @@ ## TEXT RECOGNITION -- [DATA PREPARATION](#DATA_PREPARATION) - - [Dataset Download](#Dataset_download) - - [Costom Dataset](#Costom_Dataset) - - [Dictionary](#Dictionary) - - [Add Space Category](#Add_space_category) +- [1 DATA PREPARATION](#DATA_PREPARATION) + - [1.1 Costom Dataset](#Costom_Dataset) + - [1.2 Dataset Download](#Dataset_download) + - [1.3 Dictionary](#Dictionary) + - [1.4 Add Space Category](#Add_space_category) -- [TRAINING](#TRAINING) - - [Data Augmentation](#Data_Augmentation) - - [Training](#Training) - - [Multi-language](#Multi_language) +- [2 TRAINING](#TRAINING) + - [2.1 Data Augmentation](#Data_Augmentation) + - [2.2 Training](#Training) + - [2.3 Multi-language](#Multi_language) -- [EVALUATION](#EVALUATION) +- [3 EVALUATION](#EVALUATION) -- [PREDICTION](#PREDICTION) - - [Training engine prediction](#Training_engine_prediction) +- [4 PREDICTION](#PREDICTION) + - [4.1 Training engine prediction](#Training_engine_prediction) ### DATA PREPARATION -PaddleOCR supports two data formats: `LMDB` is used to train public data and evaluation algorithms; `general data` is used to train your own data: +PaddleOCR supports two data formats: +- `LMDB` is used to train data sets stored in lmdb format; +- `general data` is used to train data sets stored in text files: Please organize the dataset as follows: The default storage path for training data is `PaddleOCR/train_data`, if you already have a dataset on your disk, just create a soft link to the dataset directory: ``` +# linux and mac os ln -sf /train_data/dataset +# windows +mklink /d /train_data/dataset ``` - -* Dataset download - -If you do not have a dataset locally, you can download it on the official website [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads). Also refer to [DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here)，download the lmdb format dataset required for benchmark - -If you want to reproduce the paper indicators of SRN, you need to download offline [augmented data](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA), extraction code: y3ry. The augmented data is obtained by rotation and perturbation of mjsynth and synthtext. Please unzip the data to {your_path}/PaddleOCR/train_data/data_lmdb_Release/training/path. - -* Use your own dataset: +#### 1.1 Costom dataset If you want to use your own data for training, please refer to the following to organize your data. - Training set -First put the training images in the same folder (train_images), and use a txt file (rec_gt_train.txt) to store the image path and label. +It is recommended to put the training images in the same folder, and use a txt file (rec_gt_train.txt) to store the image path and label. The contents of the txt file are as follows: * Note: by default, the image path and image label are split with \t, if you use other methods to split, it will cause training error ``` " Image file name Image annotation " -train_data/train_0001.jpg 简单可依赖 -train_data/train_0002.jpg 用科技让复杂的世界更简单 -``` -PaddleOCR provides label files for training the icdar2015 dataset, which can be downloaded in the following ways: - -``` -# Training set label -wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt -# Test Set Label -wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt +train_data/rec/train/word_001.jpg 简单可依赖 +train_data/rec/train/word_002.jpg 用科技让复杂的世界更简单 +... ``` The final training set should have the following file structure: ``` |-train_data - |-ic15_data - |- rec_gt_train.txt - |- train - |- word_001.png - |- word_002.jpg - |- word_003.jpg - | ... + |-rec + |- rec_gt_train.txt + |- train + |- word_001.png + |- word_002.jpg + |- word_003.jpg + | ... ``` - Test set @@ -82,6 +73,7 @@ Similar to the training set, the test set also needs to be provided a folder con ``` |-train_data + |-rec |-ic15_data |- rec_gt_test.txt |- test @@ -90,8 +82,25 @@ Similar to the training set, the test set also needs to be provided a folder con |- word_003.jpg | ... ``` + + +#### 1.2 Dataset download + +If you do not have a dataset locally, you can download it on the official website [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads). Also refer to [DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，download the lmdb format dataset required for benchmark + +If you want to reproduce the paper indicators of SRN, you need to download offline [augmented data](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA), extraction code: y3ry. The augmented data is obtained by rotation and perturbation of mjsynth and synthtext. Please unzip the data to {your_path}/PaddleOCR/train_data/data_lmdb_Release/training/path. + +PaddleOCR provides label files for training the icdar2015 dataset, which can be downloaded in the following ways: + +``` +# Training set label +wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt +# Test Set Label +wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt +``` + -- Dictionary +#### 1.3 Dictionary Finally, a dictionary ({word_dict_name}.txt) needs to be provided so that when the model is trained, all the characters that appear can be mapped to the dictionary index. @@ -108,6 +117,8 @@ n In `word_dict.txt`, there is a single word in each line, which maps characters and numeric indexes together, e.g "and" will be mapped to [2 5 1] +PaddleOCR has built-in dictionaries, which can be used on demand. + `ppocr/utils/ppocr_keys_v1.txt` is a Chinese dictionary with 6623 characters. `ppocr/utils/ic15_dict.txt` is an English dictionary with 63 characters @@ -123,10 +134,8 @@ In `word_dict.txt`, there is a single word in each line, which maps characters a `ppocr/utils/dict/en_dict.txt` is a English dictionary with 63 characters -You can use it on demand. - The current multi-language model is still in the demo stage and will continue to optimize the model and add languages. **You are very welcome to provide us with dictionaries and fonts in other languages**, -If you like, you can submit the dictionary file to [dict](../../ppocr/utils/dict) or corpus file to [corpus](../../ppocr/utils/corpus) and we will thank you in the Repo. +If you like, you can submit the dictionary file to [dict](../../ppocr/utils/dict) and we will thank you in the Repo. To customize the dict file, please modify the `character_dict_path` field in `configs/rec/rec_icdar15_train.yml` and set `character_type` to `ch`. @@ -136,14 +145,14 @@ To customize the dict file, please modify the `character_dict_path` field in `co If you need to customize dic file, please add character_dict_path field in configs/rec/rec_icdar15_train.yml to point to your dictionary path. And set character_type to ch. -- Add space category +#### 1.4 Add space category If you want to support the recognition of the `space` category, please set the `use_space_char` field in the yml file to `True`. **Note: use_space_char only takes effect when character_type=ch** -### TRAINING +### 2 TRAINING PaddleOCR provides training scripts, evaluation scripts, and prediction scripts. In this section, the CRNN recognition model will be used as an example: @@ -166,7 +175,7 @@ Start training: python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml ``` -- Data Augmentation +#### 2.1 Data Augmentation PaddleOCR provides a variety of data augmentation methods. If you want to add disturbance during training, please set `distort: true` in the configuration file. @@ -175,7 +184,7 @@ The default perturbation methods are: cvtColor, blur, jitter, Gasuss noise, rand Each disturbance method is selected with a 50% probability during the training process. For specific code implementation, please refer to: [img_tools.py](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/ppocr/data/rec/img_tools.py) -- Training +#### 2.2 Training PaddleOCR supports alternating training and evaluation. You can modify `eval_batch_step` in `configs/rec/rec_icdar15_train.yml` to set the evaluation frequency. By default, it is evaluated every 500 iter and the best acc model is saved under `output/rec_CRNN/best_accuracy` during the evaluation process. @@ -195,6 +204,10 @@ If the evaluation set is large, the test will be time-consuming. It is recommend | rec_mv3_none_none_ctc.yml | Rosetta | Mobilenet_v3 large 0.5 | None | None | ctc | | rec_r34_vd_none_bilstm_ctc.yml | CRNN | Resnet34_vd | None | BiLSTM | ctc | | rec_r34_vd_none_none_ctc.yml | Rosetta | Resnet34_vd | None | None | ctc | +| rec_mv3_tps_bilstm_att.yml | CRNN | Mobilenet_v3 | TPS | BiLSTM | att | +| rec_r34_vd_tps_bilstm_att.yml | CRNN | Resnet34_vd | TPS | BiLSTM | att | +| rec_r50fpn_vd_none_srn.yml | SRN | Resnet50_fpn_vd | None | rnn | srn | + For training Chinese data, it is recommended to use [rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml). If you want to try the result of other algorithms on the Chinese data set, please refer to the following instructions to modify the configuration file: @@ -264,7 +277,7 @@ Eval: **Note that the configuration file for prediction/evaluation must be consistent with the training.** -- Multi-language +#### 2.3 Multi-language PaddleOCR currently supports 26 (except Chinese) language recognition. A multi-language configuration file template is provided under the path `configs/rec/multi_languages`: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。 @@ -416,7 +429,7 @@ Eval: ``` -### EVALUATION +### 3 EVALUATION The evaluation dataset can be set by modifying the `Eval.dataset.label_file_list` field in the `configs/rec/rec_icdar15_train.yml` file. @@ -426,10 +439,10 @@ python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec ``` -### PREDICTION +### 4 PREDICTION -* Training engine prediction +#### 4.1 Training engine prediction Using the model trained by paddleocr, you can quickly get prediction through the following script. diff --git a/doc/doc_en/whl_en.md b/doc/doc_en/whl_en.md index 1ef14f1427eb4c1a2a504f4a420cd43c8444aeac..ae4d34923535779d89594ee7f6fa4259f38ba497 100644 --- a/doc/doc_en/whl_en.md +++ b/doc/doc_en/whl_en.md @@ -1,7 +1,7 @@ # paddleocr package -## Get started quickly -### install package +## 1 Get started quickly +### 1.1 install package install by pypi ```bash pip install "paddleocr>=2.0.1" # Recommend to use version 2.0.1+ @@ -12,9 +12,11 @@ build own whl package and install python3 setup.py bdist_wheel pip3 install dist/paddleocr-x.x.x-py3-none-any.whl # x.x.x is the version of paddleocr ``` -### 1. Use by code +## 2 Use +### 2.1 Use by code +The paddleocr whl package will automatically download the ppocr lightweight model as the default model, which can be customized and replaced according to the section 3 **Custom Model**. -* detection classification and recognition +* detection angle classification and recognition ```python from paddleocr import PaddleOCR,draw_ocr # Paddleocr supports Chinese, English, French, German, Korean and Japanese. @@ -163,7 +165,7 @@ Output will be a list, each item contains classification result and confidence ['0', 0.99999964] ``` -### Use by command line +### 2.2 Use by command line show help information ```bash @@ -239,11 +241,11 @@ Output will be a list, each item contains classification result and confidence ['0', 0.99999964] ``` -## Use custom model +## 3 Use custom model When the built-in model cannot meet the needs, you need to use your own trained model. First, refer to the first section of [inference_en.md](./inference_en.md) to convert your det and rec model to inference model, and then use it as follows -### 1. Use by code +### 3.1 Use by code ```python from paddleocr import PaddleOCR,draw_ocr @@ -265,17 +267,17 @@ im_show = Image.fromarray(im_show) im_show.save('result.jpg') ``` -### Use by command line +### 3.2 Use by command line ```bash paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --det_model_dir {your_det_model_dir} --rec_model_dir {your_rec_model_dir} --rec_char_dict_path {your_rec_char_dict_path} --cls_model_dir {your_cls_model_dir} --use_angle_cls true ``` -### Use web images or numpy array as input +## 4 Use web images or numpy array as input -1. Web image +### 4.1 Web image -Use by code +- Use by code ```python from paddleocr import PaddleOCR, draw_ocr ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory @@ -294,12 +296,12 @@ im_show = draw_ocr(image, boxes, txts, scores, font_path='/path/to/PaddleOCR/doc im_show = Image.fromarray(im_show) im_show.save('result.jpg') ``` -Use by command line +- Use by command line ```bash paddleocr --image_dir http://n.sinaimg.cn/ent/transform/w630h933/20171222/o111-fypvuqf1838418.jpg --use_angle_cls=true ``` -2. Numpy array +### 4.2 Numpy array Support numpy array as input only when used by code ```python @@ -324,7 +326,7 @@ im_show.save('result.jpg') ``` -## Parameter Description +## 5 Parameter Description | Parameter | Description | Default value | |-------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------| diff --git a/doc/imgs/model_prod_flow_ch.png b/doc/imgs/model_prod_flow_ch.png new file mode 100644 index 0000000000000000000000000000000000000000..4906b2716e2b3cf278416e1766b8a63329362693 Binary files /dev/null and b/doc/imgs/model_prod_flow_ch.png differ diff --git a/doc/imgs_en/model_prod_flow_en.png b/doc/imgs_en/model_prod_flow_en.png new file mode 100644 index 0000000000000000000000000000000000000000..eb5cb9cc9fe065cf2d1974ae3a0d7de12f8bc65d Binary files /dev/null and b/doc/imgs_en/model_prod_flow_en.png differ diff --git a/doc/imgs_results/angle_class_example.jpg b/doc/imgs_results/angle_class_example.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8e683be32cdb20e964a7154980d5b1d33d6a8eca Binary files /dev/null and b/doc/imgs_results/angle_class_example.jpg differ diff --git a/paddleocr.py b/paddleocr.py index db24aa59e9237ce9cafa972673ecb0b1a3357c33..7c126261eff1168a1888d72f71fb284e347f9ec9 100644 --- a/paddleocr.py +++ b/paddleocr.py @@ -146,7 +146,8 @@ def parse_args(mMain=True, add_help=True): # DB parmas parser.add_argument("--det_db_thresh", type=float, default=0.3) parser.add_argument("--det_db_box_thresh", type=float, default=0.5) - parser.add_argument("--det_db_unclip_ratio", type=float, default=2.0) + parser.add_argument("--det_db_unclip_ratio", type=float, default=1.6) + parser.add_argument("--use_dilation", type=bool, default=False) # EAST parmas parser.add_argument("--det_east_score_thresh", type=float, default=0.8) @@ -193,7 +194,8 @@ def parse_args(mMain=True, add_help=True): det_limit_type='max', det_db_thresh=0.3, det_db_box_thresh=0.5, - det_db_unclip_ratio=2.0, + det_db_unclip_ratio=1.6, + use_dilation=False, det_east_score_thresh=0.8, det_east_cover_thresh=0.1, det_east_nms_thresh=0.2, diff --git a/ppocr/data/imaug/label_ops.py b/ppocr/data/imaug/label_ops.py index 61c0c196b3a48911707dc5210a410145ec93a76d..7a32d870bfc7f532896ce6b11aac5508a6369993 100644 --- a/ppocr/data/imaug/label_ops.py +++ b/ppocr/data/imaug/label_ops.py @@ -199,16 +199,30 @@ class AttnLabelEncode(BaseRecLabelEncode): super(AttnLabelEncode, self).__init__(max_text_length, character_dict_path, character_type, use_space_char) - self.beg_str = "sos" - self.end_str = "eos" def add_special_char(self, dict_character): - dict_character = [self.beg_str, self.end_str] + dict_character + self.beg_str = "sos" + self.end_str = "eos" + dict_character = [self.beg_str] + dict_character + [self.end_str] return dict_character - def __call__(self, text): + def __call__(self, data): + text = data['label'] text = self.encode(text) - return text + if text is None: + return None + if len(text) >= self.max_text_len: + return None + data['length'] = np.array(len(text)) + text = [0] + text + [len(self.character) - 1] + [0] * (self.max_text_len + - len(text) - 2) + data['label'] = np.array(text) + return data + + def get_ignored_tokens(self): + beg_idx = self.get_beg_end_flag_idx("beg") + end_idx = self.get_beg_end_flag_idx("end") + return [beg_idx, end_idx] def get_beg_end_flag_idx(self, beg_or_end): if beg_or_end == "beg": @@ -241,13 +255,13 @@ class SRNLabelEncode(BaseRecLabelEncode): def __call__(self, data): text = data['label'] text = self.encode(text) - char_num = len(self.character_str) + char_num = len(self.character) if text is None: return None if len(text) > self.max_text_len: return None data['length'] = np.array(len(text)) - text = text + [char_num] * (self.max_text_len - len(text)) + text = text + [char_num - 1] * (self.max_text_len - len(text)) data['label'] = np.array(text) return data diff --git a/ppocr/data/imaug/make_shrink_map.py b/ppocr/data/imaug/make_shrink_map.py index a66706f2bff25680a3597ef67ce5fa809747cf05..03385e3e10b3970cb74744e78632f97384281607 100644 --- a/ppocr/data/imaug/make_shrink_map.py +++ b/ppocr/data/imaug/make_shrink_map.py @@ -44,21 +44,34 @@ class MakeShrinkMap(object): ignore_tags[i] = True else: polygon_shape = Polygon(polygon) - distance = polygon_shape.area * ( - 1 - np.power(self.shrink_ratio, 2)) / polygon_shape.length - subject = [tuple(l) for l in text_polys[i]] + subject = [tuple(l) for l in polygon] padding = pyclipper.PyclipperOffset() padding.AddPath(subject, pyclipper.JT_ROUND, pyclipper.ET_CLOSEDPOLYGON) - shrinked = padding.Execute(-distance) + shrinked = [] + + # Increase the shrink ratio every time we get multiple polygon returned back + possible_ratios = np.arange(self.shrink_ratio, 1, + self.shrink_ratio) + np.append(possible_ratios, 1) + # print(possible_ratios) + for ratio in possible_ratios: + # print(f"Change shrink ratio to {ratio}") + distance = polygon_shape.area * ( + 1 - np.power(ratio, 2)) / polygon_shape.length + shrinked = padding.Execute(-distance) + if len(shrinked) == 1: + break + if shrinked == []: cv2.fillPoly(mask, polygon.astype(np.int32)[np.newaxis, :, :], 0) ignore_tags[i] = True continue - shrinked = np.array(shrinked[0]).reshape(-1, 2) - cv2.fillPoly(gt, [shrinked.astype(np.int32)], 1) - # cv2.fillPoly(gt[0], [shrinked.astype(np.int32)], 1) + + for each_shirnk in shrinked: + shirnk = np.array(each_shirnk).reshape(-1, 2) + cv2.fillPoly(gt, [shirnk.astype(np.int32)], 1) data['shrink_map'] = gt data['shrink_mask'] = mask @@ -84,11 +97,12 @@ class MakeShrinkMap(object): return polygons, ignore_tags def polygon_area(self, polygon): - # return cv2.contourArea(polygon.astype(np.float32)) - edge = 0 - for i in range(polygon.shape[0]): - next_index = (i + 1) % polygon.shape[0] - edge += (polygon[next_index, 0] - polygon[i, 0]) * ( - polygon[next_index, 1] - polygon[i, 1]) - - return edge / 2. + """ + compute polygon area + """ + area = 0 + q = polygon[-1] + for p in polygon: + area += p[0] * q[1] - p[1] * q[0] + q = p + return area / 2.0 diff --git a/ppocr/data/imaug/operators.py b/ppocr/data/imaug/operators.py index 8b9175cf0b201969fdfbd7fb082200e2241f4444..eacfdf3b243af5b9051ad726ced6edacddff45ed 100644 --- a/ppocr/data/imaug/operators.py +++ b/ppocr/data/imaug/operators.py @@ -185,8 +185,8 @@ class DetResizeForTest(object): resize_h = int(h * ratio) resize_w = int(w * ratio) - resize_h = int(round(resize_h / 32) * 32) - resize_w = int(round(resize_w / 32) * 32) + resize_h = max(int(round(resize_h / 32) * 32), 32) + resize_w = max(int(round(resize_w / 32) * 32), 32) try: if int(resize_w) <= 0 or int(resize_h) <= 0: diff --git a/ppocr/losses/__init__.py b/ppocr/losses/__init__.py index b280eb333e8910ac2378962eb4ffde5a98f31efd..3881abf7741b8be78306bd070afb11df15606327 100755 --- a/ppocr/losses/__init__.py +++ b/ppocr/losses/__init__.py @@ -23,13 +23,15 @@ def build_loss(config): # rec loss from .rec_ctc_loss import CTCLoss + from .rec_att_loss import AttentionLoss from .rec_srn_loss import SRNLoss # cls loss from .cls_loss import ClsLoss support_dict = [ - 'DBLoss', 'EASTLoss', 'SASTLoss', 'CTCLoss', 'ClsLoss', 'SRNLoss' + 'DBLoss', 'EASTLoss', 'SASTLoss', 'CTCLoss', 'ClsLoss', 'AttentionLoss', + 'SRNLoss' ] config = copy.deepcopy(config) diff --git a/ppocr/losses/det_basic_loss.py b/ppocr/losses/det_basic_loss.py index 57b3667d9f32a871f748c40a65429551613991ca..eba5526dd2bd1c0328130b50817172df437cc360 100644 --- a/ppocr/losses/det_basic_loss.py +++ b/ppocr/losses/det_basic_loss.py @@ -200,6 +200,6 @@ def ohem_batch(scores, gt_texts, training_masks, ohem_ratio): i, :, :], ohem_ratio)) selected_masks = np.concatenate(selected_masks, 0) - selected_masks = paddle.to_variable(selected_masks) + selected_masks = paddle.to_tensor(selected_masks) return selected_masks diff --git a/ppocr/losses/rec_att_loss.py b/ppocr/losses/rec_att_loss.py new file mode 100644 index 0000000000000000000000000000000000000000..6e2f67483c86a45f3aa1feb1e1fac1a5013bfb46 --- /dev/null +++ b/ppocr/losses/rec_att_loss.py @@ -0,0 +1,39 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from paddle import nn + + +class AttentionLoss(nn.Layer): + def __init__(self, **kwargs): + super(AttentionLoss, self).__init__() + self.loss_func = nn.CrossEntropyLoss(weight=None, reduction='none') + + def forward(self, predicts, batch): + targets = batch[1].astype("int64") + label_lengths = batch[2].astype('int64') + batch_size, num_steps, num_classes = predicts.shape[0], predicts.shape[ + 1], predicts.shape[2] + assert len(targets.shape) == len(list(predicts.shape)) - 1, \ + "The target's shape and inputs's shape is [N, d] and [N, num_steps]" + + inputs = paddle.reshape(predicts, [-1, predicts.shape[-1]]) + targets = paddle.reshape(targets, [-1]) + + return {'loss': paddle.sum(self.loss_func(inputs, targets))} diff --git a/ppocr/metrics/rec_metric.py b/ppocr/metrics/rec_metric.py index b3aa9f38f8378eee7104d7e3696b86bede0de903..66c084d771dece0e2974bc72a177b53f564a8f2e 100644 --- a/ppocr/metrics/rec_metric.py +++ b/ppocr/metrics/rec_metric.py @@ -29,7 +29,7 @@ class RecMetric(object): pred = pred.replace(" ", "") target = target.replace(" ", "") norm_edit_dis += Levenshtein.distance(pred, target) / max( - len(pred), len(target)) + len(pred), len(target), 1) if pred == target: correct_num += 1 all_num += 1 diff --git a/ppocr/modeling/heads/__init__.py b/ppocr/modeling/heads/__init__.py index 1a39ca412a1faf9a8cefb1de0db66c33ed9dc27e..efe05718506e94a5ae8ad5ff47bcff26d44c1473 100755 --- a/ppocr/modeling/heads/__init__.py +++ b/ppocr/modeling/heads/__init__.py @@ -23,12 +23,14 @@ def build_head(config): # rec head from .rec_ctc_head import CTCHead + from .rec_att_head import AttentionHead from .rec_srn_head import SRNHead # cls head from .cls_head import ClsHead support_dict = [ - 'DBHead', 'EASTHead', 'SASTHead', 'CTCHead', 'ClsHead', 'SRNHead' + 'DBHead', 'EASTHead', 'SASTHead', 'CTCHead', 'ClsHead', 'AttentionHead', + 'SRNHead' ] module_name = config.pop('name') diff --git a/ppocr/modeling/heads/rec_att_head.py b/ppocr/modeling/heads/rec_att_head.py new file mode 100644 index 0000000000000000000000000000000000000000..0d222714ff7edebfc717daa81d48ce7424dfbd03 --- /dev/null +++ b/ppocr/modeling/heads/rec_att_head.py @@ -0,0 +1,202 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +import numpy as np + + +class AttentionHead(nn.Layer): + def __init__(self, in_channels, out_channels, hidden_size, **kwargs): + super(AttentionHead, self).__init__() + self.input_size = in_channels + self.hidden_size = hidden_size + self.num_classes = out_channels + + self.attention_cell = AttentionGRUCell( + in_channels, hidden_size, out_channels, use_gru=False) + self.generator = nn.Linear(hidden_size, out_channels) + + def _char_to_onehot(self, input_char, onehot_dim): + input_ont_hot = F.one_hot(input_char, onehot_dim) + return input_ont_hot + + def forward(self, inputs, targets=None, batch_max_length=25): + batch_size = inputs.shape[0] + num_steps = batch_max_length + + hidden = paddle.zeros((batch_size, self.hidden_size)) + output_hiddens = [] + + if targets is not None: + for i in range(num_steps): + char_onehots = self._char_to_onehot( + targets[:, i], onehot_dim=self.num_classes) + (outputs, hidden), alpha = self.attention_cell(hidden, inputs, + char_onehots) + output_hiddens.append(paddle.unsqueeze(outputs, axis=1)) + output = paddle.concat(output_hiddens, axis=1) + probs = self.generator(output) + + else: + targets = paddle.zeros(shape=[batch_size], dtype="int32") + probs = None + char_onehots = None + outputs = None + alpha = None + + for i in range(num_steps): + char_onehots = self._char_to_onehot( + targets, onehot_dim=self.num_classes) + (outputs, hidden), alpha = self.attention_cell(hidden, inputs, + char_onehots) + probs_step = self.generator(outputs) + if probs is None: + probs = paddle.unsqueeze(probs_step, axis=1) + else: + probs = paddle.concat( + [probs, paddle.unsqueeze( + probs_step, axis=1)], axis=1) + next_input = probs_step.argmax(axis=1) + targets = next_input + + return probs + + +class AttentionGRUCell(nn.Layer): + def __init__(self, input_size, hidden_size, num_embeddings, use_gru=False): + super(AttentionGRUCell, self).__init__() + self.i2h = nn.Linear(input_size, hidden_size, bias_attr=False) + self.h2h = nn.Linear(hidden_size, hidden_size) + self.score = nn.Linear(hidden_size, 1, bias_attr=False) + + self.rnn = nn.GRUCell( + input_size=input_size + num_embeddings, hidden_size=hidden_size) + + self.hidden_size = hidden_size + + def forward(self, prev_hidden, batch_H, char_onehots): + + batch_H_proj = self.i2h(batch_H) + prev_hidden_proj = paddle.unsqueeze(self.h2h(prev_hidden), axis=1) + + res = paddle.add(batch_H_proj, prev_hidden_proj) + res = paddle.tanh(res) + e = self.score(res) + + alpha = F.softmax(e, axis=1) + alpha = paddle.transpose(alpha, [0, 2, 1]) + context = paddle.squeeze(paddle.mm(alpha, batch_H), axis=1) + concat_context = paddle.concat([context, char_onehots], 1) + + cur_hidden = self.rnn(concat_context, prev_hidden) + + return cur_hidden, alpha + + +class AttentionLSTM(nn.Layer): + def __init__(self, in_channels, out_channels, hidden_size, **kwargs): + super(AttentionLSTM, self).__init__() + self.input_size = in_channels + self.hidden_size = hidden_size + self.num_classes = out_channels + + self.attention_cell = AttentionLSTMCell( + in_channels, hidden_size, out_channels, use_gru=False) + self.generator = nn.Linear(hidden_size, out_channels) + + def _char_to_onehot(self, input_char, onehot_dim): + input_ont_hot = F.one_hot(input_char, onehot_dim) + return input_ont_hot + + def forward(self, inputs, targets=None, batch_max_length=25): + batch_size = inputs.shape[0] + num_steps = batch_max_length + + hidden = (paddle.zeros((batch_size, self.hidden_size)), paddle.zeros( + (batch_size, self.hidden_size))) + output_hiddens = [] + + if targets is not None: + for i in range(num_steps): + # one-hot vectors for a i-th char + char_onehots = self._char_to_onehot( + targets[:, i], onehot_dim=self.num_classes) + hidden, alpha = self.attention_cell(hidden, inputs, + char_onehots) + + hidden = (hidden[1][0], hidden[1][1]) + output_hiddens.append(paddle.unsqueeze(hidden[0], axis=1)) + output = paddle.concat(output_hiddens, axis=1) + probs = self.generator(output) + + else: + targets = paddle.zeros(shape=[batch_size], dtype="int32") + probs = None + + for i in range(num_steps): + char_onehots = self._char_to_onehot( + targets, onehot_dim=self.num_classes) + hidden, alpha = self.attention_cell(hidden, inputs, + char_onehots) + probs_step = self.generator(hidden[0]) + hidden = (hidden[1][0], hidden[1][1]) + if probs is None: + probs = paddle.unsqueeze(probs_step, axis=1) + else: + probs = paddle.concat( + [probs, paddle.unsqueeze( + probs_step, axis=1)], axis=1) + + next_input = probs_step.argmax(axis=1) + + targets = next_input + + return probs + + +class AttentionLSTMCell(nn.Layer): + def __init__(self, input_size, hidden_size, num_embeddings, use_gru=False): + super(AttentionLSTMCell, self).__init__() + self.i2h = nn.Linear(input_size, hidden_size, bias_attr=False) + self.h2h = nn.Linear(hidden_size, hidden_size) + self.score = nn.Linear(hidden_size, 1, bias_attr=False) + if not use_gru: + self.rnn = nn.LSTMCell( + input_size=input_size + num_embeddings, hidden_size=hidden_size) + else: + self.rnn = nn.GRUCell( + input_size=input_size + num_embeddings, hidden_size=hidden_size) + + self.hidden_size = hidden_size + + def forward(self, prev_hidden, batch_H, char_onehots): + batch_H_proj = self.i2h(batch_H) + prev_hidden_proj = paddle.unsqueeze(self.h2h(prev_hidden[0]), axis=1) + res = paddle.add(batch_H_proj, prev_hidden_proj) + res = paddle.tanh(res) + e = self.score(res) + + alpha = F.softmax(e, axis=1) + alpha = paddle.transpose(alpha, [0, 2, 1]) + context = paddle.squeeze(paddle.mm(alpha, batch_H), axis=1) + concat_context = paddle.concat([context, char_onehots], 1) + cur_hidden = self.rnn(concat_context, prev_hidden) + + return cur_hidden, alpha diff --git a/ppocr/modeling/heads/rec_srn_head.py b/ppocr/modeling/heads/rec_srn_head.py index 8aaf65e1ae018dd410bbc05d0d7dcac821d062a3..d2c7fc028d28c79057708d4e6f306c417ba6306a 100644 --- a/ppocr/modeling/heads/rec_srn_head.py +++ b/ppocr/modeling/heads/rec_srn_head.py @@ -246,7 +246,7 @@ class SRNHead(nn.Layer): num_encoder_tus=self.num_encoder_TUs, num_decoder_tus=self.num_decoder_TUs, hidden_dims=self.hidden_dims) - self.vsfd = VSFD(in_channels=in_channels) + self.vsfd = VSFD(in_channels=in_channels, char_num=self.char_num) self.gsrm.wrap_encoder1.prepare_decoder.emb0 = self.gsrm.wrap_encoder0.prepare_decoder.emb0 diff --git a/ppocr/postprocess/rec_postprocess.py b/ppocr/postprocess/rec_postprocess.py index 76a700e1599b143288814778dcc948126a98151d..af243caa44e8390657b7a95e971aede0c0f90edd 100644 --- a/ppocr/postprocess/rec_postprocess.py +++ b/ppocr/postprocess/rec_postprocess.py @@ -135,16 +135,62 @@ class AttnLabelDecode(BaseRecLabelDecode): **kwargs): super(AttnLabelDecode, self).__init__(character_dict_path, character_type, use_space_char) - self.beg_str = "sos" - self.end_str = "eos" def add_special_char(self, dict_character): - dict_character = [self.beg_str, self.end_str] + dict_character + self.beg_str = "sos" + self.end_str = "eos" + dict_character = dict_character + dict_character = [self.beg_str] + dict_character + [self.end_str] return dict_character - def __call__(self, text): + def decode(self, text_index, text_prob=None, is_remove_duplicate=False): + """ convert text-index into text-label. """ + result_list = [] + ignored_tokens = self.get_ignored_tokens() + [beg_idx, end_idx] = self.get_ignored_tokens() + batch_size = len(text_index) + for batch_idx in range(batch_size): + char_list = [] + conf_list = [] + for idx in range(len(text_index[batch_idx])): + if text_index[batch_idx][idx] in ignored_tokens: + continue + if int(text_index[batch_idx][idx]) == int(end_idx): + break + if is_remove_duplicate: + # only for predict + if idx > 0 and text_index[batch_idx][idx - 1] == text_index[ + batch_idx][idx]: + continue + char_list.append(self.character[int(text_index[batch_idx][ + idx])]) + if text_prob is not None: + conf_list.append(text_prob[batch_idx][idx]) + else: + conf_list.append(1) + text = ''.join(char_list) + result_list.append((text, np.mean(conf_list))) + return result_list + + def __call__(self, preds, label=None, *args, **kwargs): + """ text = self.decode(text) - return text + if label is None: + return text + else: + label = self.decode(label, is_remove_duplicate=False) + return text, label + """ + if isinstance(preds, paddle.Tensor): + preds = preds.numpy() + + preds_idx = preds.argmax(axis=2) + preds_prob = preds.max(axis=2) + text = self.decode(preds_idx, preds_prob, is_remove_duplicate=False) + if label is None: + return text + label = self.decode(label, is_remove_duplicate=False) + return text, label def get_ignored_tokens(self): beg_idx = self.get_beg_end_flag_idx("beg") diff --git a/requirements.txt b/requirements.txt index 132189634919760156b421570284faa2c6fc957f..2401d52b48c10bad5ea5b244a0fd4c4365b94f09 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,5 +1,6 @@ shapely -imgaug +scikit-image==0.17.2 +imgaug==0.4.0 pyclipper lmdb opencv-python==4.2.0.32 diff --git a/tools/eval.py b/tools/eval.py index 16cfe532aae49ce98bc9503ca73e009bf206caa7..4afed469c875ef8d2200cdbfd89e5a8af4c6b7c3 100755 --- a/tools/eval.py +++ b/tools/eval.py @@ -47,6 +47,7 @@ def main(): config['Architecture']["Head"]['out_channels'] = len( getattr(post_process_class, 'character')) model = build_model(config['Architecture']) + use_srn = config['Architecture']['algorithm'] == "SRN" best_model_dict = init_model(config, model, logger) if len(best_model_dict): @@ -59,7 +60,7 @@ def main(): # start eval metirc = program.eval(model, valid_dataloader, post_process_class, - eval_class) + eval_class, use_srn) logger.info('metric eval ***************') for k, v in metirc.items(): logger.info('{}:{}'.format(k, v)) diff --git a/tools/infer/predict_det.py b/tools/infer/predict_det.py index 077692afa84a745cb1b1fcb5b2c71f3dd5653013..76c6a4478ae37e650901c8b8704e15e4a94911c4 100755 --- a/tools/infer/predict_det.py +++ b/tools/infer/predict_det.py @@ -64,7 +64,7 @@ class TextDetector(object): postprocess_params["box_thresh"] = args.det_db_box_thresh postprocess_params["max_candidates"] = 1000 postprocess_params["unclip_ratio"] = args.det_db_unclip_ratio - postprocess_params["use_dilation"] = True + postprocess_params["use_dilation"] = args.use_dilation elif self.det_algorithm == "EAST": postprocess_params['name'] = 'EASTPostProcess' postprocess_params["score_thresh"] = args.det_east_score_thresh diff --git a/tools/infer/predict_rec.py b/tools/infer/predict_rec.py index fd895e50719941877fd620cab929a20c7d88b8e5..b24e57dd973bc0216f2875232bcec6e36ab47e29 100755 --- a/tools/infer/predict_rec.py +++ b/tools/infer/predict_rec.py @@ -54,6 +54,13 @@ class TextRecognizer(object): "character_dict_path": args.rec_char_dict_path, "use_space_char": args.use_space_char } + elif self.rec_algorithm == "RARE": + postprocess_params = { + 'name': 'AttnLabelDecode', + "character_type": args.rec_char_type, + "character_dict_path": args.rec_char_dict_path, + "use_space_char": args.use_space_char + } self.postprocess_op = build_post_process(postprocess_params) self.predictor, self.input_tensor, self.output_tensors = \ utility.create_predictor(args, 'rec', logger) @@ -241,9 +248,11 @@ class TextRecognizer(object): def main(args): image_file_list = get_image_file_list(args.image_dir) text_recognizer = TextRecognizer(args) + total_run_time = 0.0 + total_images_num = 0 valid_image_file_list = [] img_list = [] - for image_file in image_file_list: + for idx, image_file in enumerate(image_file_list): img, flag = check_and_read_gif(image_file) if not flag: img = cv2.imread(image_file) @@ -252,22 +261,29 @@ def main(args): continue valid_image_file_list.append(image_file) img_list.append(img) - try: - rec_res, predict_time = text_recognizer(img_list) - except: - logger.info(traceback.format_exc()) - logger.info( - "ERROR!!!! \n" - "Please read the FAQ：https://github.com/PaddlePaddle/PaddleOCR#faq \n" - "If your model has tps module: " - "TPS does not support variable shape.\n" - "Please set --rec_image_shape='3,32,100' and --rec_char_type='en' ") - exit() - for ino in range(len(img_list)): - logger.info("Predicts of {}:{}".format(valid_image_file_list[ino], - rec_res[ino])) + if len(img_list) >= args.rec_batch_num or idx == len( + image_file_list) - 1: + try: + rec_res, predict_time = text_recognizer(img_list) + total_run_time += predict_time + except: + logger.info(traceback.format_exc()) + logger.info( + "ERROR!!!! \n" + "Please read the FAQ：https://github.com/PaddlePaddle/PaddleOCR#faq \n" + "If your model has tps module: " + "TPS does not support variable shape.\n" + "Please set --rec_image_shape='3,32,100' and --rec_char_type='en' " + ) + exit() + for ino in range(len(img_list)): + logger.info("Predicts of {}:{}".format(valid_image_file_list[ + ino], rec_res[ino])) + total_images_num += len(valid_image_file_list) + valid_image_file_list = [] + img_list = [] logger.info("Total predict time for {} images, cost: {:.3f}".format( - len(img_list), predict_time)) + total_images_num, total_run_time)) if __name__ == "__main__": diff --git a/tools/infer/predict_system.py b/tools/infer/predict_system.py index 8c4f9214db9621fe4e0393ed3dac0e9a7ccedbf6..de7ee9d342063161f2e329c99d2428051c0ecf8c 100755 --- a/tools/infer/predict_system.py +++ b/tools/infer/predict_system.py @@ -184,4 +184,4 @@ def main(args): if __name__ == "__main__": - main(utility.parse_args()) \ No newline at end of file + main(utility.parse_args()) diff --git a/tools/infer/utility.py b/tools/infer/utility.py index 4171a29bdd4194813638b72f0aae015da48fbcb1..a4a91efdd2ec04e2a2959c77a444549ad413c13d 100755 --- a/tools/infer/utility.py +++ b/tools/infer/utility.py @@ -47,6 +47,7 @@ def parse_args(): parser.add_argument("--det_db_box_thresh", type=float, default=0.5) parser.add_argument("--det_db_unclip_ratio", type=float, default=1.6) parser.add_argument("--max_batch_size", type=int, default=10) + parser.add_argument("--use_dilation", type=bool, default=False) # EAST parmas parser.add_argument("--det_east_score_thresh", type=float, default=0.8) parser.add_argument("--det_east_cover_thresh", type=float, default=0.1) @@ -123,6 +124,8 @@ def create_predictor(args, mode, logger): # cache 10 different shapes for mkldnn to avoid memory leak config.set_mkldnn_cache_capacity(10) config.enable_mkldnn() + # TODO LDOUBLEV: fix mkldnn bug when bach_size > 1 + #config.set_mkldnn_op({'conv2d', 'depthwise_conv2d', 'pool2d', 'batch_norm'}) args.rec_batch_num = 1 # config.enable_memory_optim() diff --git a/tools/program.py b/tools/program.py index 694d64152f05ffd5e9329885149891f75a98ed84..ae6491768ceff0379c917c45fd29b30514eba9c1 100755 --- a/tools/program.py +++ b/tools/program.py @@ -163,6 +163,11 @@ def train(config, if type(eval_batch_step) == list and len(eval_batch_step) >= 2: start_eval_step = eval_batch_step[0] eval_batch_step = eval_batch_step[1] + if len(valid_dataloader) == 0: + logger.info( + 'No Images in eval dataset, evaluation during training will be disabled' + ) + start_eval_step = 1e111 logger.info( "During the training process, after the {}th iteration, an evaluation is run every {} iterations". format(start_eval_step, eval_batch_step)) @@ -177,6 +182,8 @@ def train(config, model_average = False model.train() + use_srn = config['Architecture']['algorithm'] == "SRN" + if 'start_epoch' in best_model_dict: start_epoch = best_model_dict['start_epoch'] else: @@ -195,7 +202,7 @@ def train(config, break lr = optimizer.get_lr() images = batch[0] - if config['Architecture']['algorithm'] == "SRN": + if use_srn: others = batch[-4:] preds = model(images, others) model_average = True @@ -222,8 +229,8 @@ def train(config, batch = [item.numpy() for item in batch] post_result = post_process_class(preds, batch[1]) eval_class(post_result, batch) - metirc = eval_class.get_metric() - train_stats.update(metirc) + metric = eval_class.get_metric() + train_stats.update(metric) if vdl_writer is not None and dist.get_rank() == 0: for k, v in train_stats.get().items(): @@ -251,8 +258,12 @@ def train(config, min_average_window=10000, max_average_window=15625) Model_Average.apply() - cur_metirc = eval(model, valid_dataloader, post_process_class, - eval_class) + cur_metric = eval( + model, + valid_dataloader, + post_process_class, + eval_class, + use_srn=use_srn) cur_metric_str = 'cur metric, {}'.format(', '.join( ['{}: {}'.format(k, v) for k, v in cur_metric.items()])) logger.info(cur_metric_str) @@ -316,7 +327,8 @@ def train(config, return -def eval(model, valid_dataloader, post_process_class, eval_class): +def eval(model, valid_dataloader, post_process_class, eval_class, + use_srn=False): model.eval() with paddle.no_grad(): total_frame = 0.0 @@ -326,9 +338,13 @@ def eval(model, valid_dataloader, post_process_class, eval_class): if idx >= len(valid_dataloader): break images = batch[0] - others = batch[-4:] start = time.time() - preds = model(images, others) + + if use_srn: + others = batch[-4:] + preds = model(images, others) + else: + preds = model(images) batch = [item.numpy() for item in batch] # Obtain usable results from post-processing methods @@ -378,6 +394,7 @@ def preprocess(is_train=False): logger = get_logger(name='root', log_file=log_file) if config['Global']['use_visualdl']: from visualdl import LogWriter + save_model_dir = config['Global']['save_model_dir'] vdl_writer_path = '{}/vdl/'.format(save_model_dir) os.makedirs(vdl_writer_path, exist_ok=True) vdl_writer = LogWriter(logdir=vdl_writer_path) diff --git a/tools/train.py b/tools/train.py index 383f8d83919b054999f19be1490b92e3d90d7eee..fab10b6437cfe5ad30bcb49d7b1b884ba9665a79 100755 --- a/tools/train.py +++ b/tools/train.py @@ -50,6 +50,12 @@ def main(config, device, logger, vdl_writer): # build dataloader train_dataloader = build_dataloader(config, 'Train', device, logger) + if len(train_dataloader) == 0: + logger.error( + 'No Images in train dataset, please check annotation file and path in the configuration file' + ) + return + if config['Eval']: valid_dataloader = build_dataloader(config, 'Eval', device, logger) else: diff --git a/train.sh b/train.sh index 8fe861a3d79d38929fc4a4f4464187f77d27ff2f..4225470cb9f545b874e5f806af22405895e8f6c7 100644 --- a/train.sh +++ b/train.sh @@ -1,2 +1,2 @@ # recommended paddle.__version__ == 2.0.0 -python3 -m paddle.distributed.launch --gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/rec/rec_mv3_none_bilstm_ctc.yml +python3 -m paddle.distributed.launch --log_dir=./debug/ --gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/rec/rec_mv3_none_bilstm_ctc.yml