diff --git a/README.md b/README.md
index 5eb9ed8fb43f132511e82bd56fa9831eb407c9a8..2035ea2e473871eb60c091c31d7de7e1d6611420 100644
--- a/README.md
+++ b/README.md
@@ -246,7 +246,8 @@ PaddleOCR文本识别算法的训练和使用请参考文档教程中[文本识
 ## 许可证书
 本项目的发布受<a href="https://github.com/PaddlePaddle/PaddleOCR/blob/master/LICENSE">Apache 2.0 license</a>许可认证。
 
-## 如何贡献代码
+## 贡献代码
 我们非常欢迎你为PaddleOCR贡献代码，也十分感谢你的反馈。
 
 - 非常感谢 [Khanh Tran](https://github.com/xxxpsyduck) 贡献了英文文档。
+- 非常感谢 [zhangxin](https://github.com/ZhangXinNan)([Blog](https://blog.csdn.net/sdlypyzq)) 贡献新的可视化方式、添加.gitgnore、处理手动设置PYTHONPATH环境变量的问题
diff --git a/README_en.md b/README_en.md
index c64b65c4cb28d5010d340e5bc1f7605c437a2300..219029968a0e950513cf9135004b2de5203d7117 100644
--- a/README_en.md
+++ b/README_en.md
@@ -255,3 +255,4 @@ This project is released under <a href="https://github.com/PaddlePaddle/PaddleOC
 We welcome all the contributions to PaddleOCR and appreciate for your feedback very much.
 
 - Many thanks to [Khanh Tran](https://github.com/xxxpsyduck) for contributing the English documentation.
+- Many thanks to [zhangxin](https://github.com/ZhangXinNan) for contributing the new visualize function、add .gitgnore and discard set PYTHONPATH manually.
diff --git a/doc/doc_ch/detection.md b/doc/doc_ch/detection.md
index fbff707555d861adf41519a15406647efbf24939..cfff12f976f0a0abe06c37187755d78358e5ca25 100644
--- a/doc/doc_ch/detection.md
+++ b/doc/doc_ch/detection.md
@@ -23,7 +23,7 @@ wget -P ./train_data/  https://paddleocr.bj.bcebos.com/dataset/test_icdar2015_la
   └─ test_icdar2015_label.txt     icdar数据集的测试标注
 ```
 
-提供的标注文件格式为：
+提供的标注文件格式为，其中中间是"\t"分隔：
 ```
 " 图像文件名                    json.dumps编码的图像标注信息"
 ch4_test_images/img_61.jpg    [{"transcription": "MASA", "points": [[310, 104], [416, 141], [418, 216], [312, 179]], ...}]
@@ -35,7 +35,7 @@ json.dumps编码前的图像标注信息是包含多个字典的list，字典中
 
 ## 快速启动训练
 
-首先下载pretrain model，PaddleOCR的检测模型目前支持两种backbone，分别是MobileNetV3、ResNet50_vd，
+首先下载模型backbone的pretrain model，PaddleOCR的检测模型目前支持两种backbone，分别是MobileNetV3、ResNet50_vd，
 您可以根据需求使用[PaddleClas](https://github.com/PaddlePaddle/PaddleClas/tree/master/ppcls/modeling/architectures)中的模型更换backbone。
 ```
 cd PaddleOCR/
@@ -62,7 +62,7 @@ tar xf ./pretrain_models/MobileNetV3_large_x0_5_pretrained.tar ./pretrain_models
 *如果您安装的是cpu版本，请将配置文件中的 `use_gpu` 字段修改为false*
 
 ```
-python3 tools/train.py -c configs/det/det_mv3_db.yml
+python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained/
 ```
 
 上述指令中，通过-c 选择训练使用configs/det/det_db_mv3.yml配置文件。
@@ -73,6 +73,15 @@ python3 tools/train.py -c configs/det/det_mv3_db.yml
 python3 tools/train.py -c configs/det/det_mv3_db.yml -o Optimizer.base_lr=0.0001
 ```
 
+**断点训练**
+
+如果训练程序中断，如果希望加载训练中断的模型从而恢复训练，可以通过指定Global.checkpoints指定要加载的模型路径：
+```
+python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=./your/trained/model
+```
+
+**注意**：Global.checkpoints的优先级高于Global.pretrain_weights的优先级，即同时指定两个参数时，优先加载Global.checkpoints指定的模型，如果Global.checkpoints指定的模型路径有误，会加载Global.pretrain_weights指定的模型。
+
 ## 指标评估
 
 PaddleOCR计算三个OCR检测相关的指标，分别是：Precision、Recall、Hmean。
diff --git a/doc/doc_ch/inference.md b/doc/doc_ch/inference.md
index 8e1fa9ffd040bb7075f8cef456a69627f48b6870..ea4bc75642311aca5a98ac66dce13e78a90e3c1a 100644
--- a/doc/doc_ch/inference.md
+++ b/doc/doc_ch/inference.md
@@ -192,6 +192,13 @@ self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
 dict_character = list(self.character_str)
 ```
 
+### 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
+
+```
+python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./your inference model" --rec_image_shape="3, 32, 100" --rec_char_type="en" --rec_char_dict_path="your text dict path"
+```
+
 ## 文本检测、识别串联推理
 
 ### 1.超轻量中文OCR模型推理
diff --git a/doc/doc_en/detection_en.md b/doc/doc_en/detection_en.md
index 619f2c01c416c007e1468607924af3046bb8a52d..7dc1fa200e24194a704cdd330b74bc38a5fe180e 100644
--- a/doc/doc_en/detection_en.md
+++ b/doc/doc_en/detection_en.md
@@ -69,6 +69,17 @@ You can also use the `-o` parameter to change the training parameters without mo
 python3 tools/train.py -c configs/det/det_mv3_db.yml -o Optimizer.base_lr=0.0001
 ```
 
+**load trained model and conntinue training**
+If you expect to load trained model and continue the training again, you can specify the `Global.checkpoints` parameter as the model path to be loaded.
+
+For example:
+```
+python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=./your/trained/model
+```
+
+**Note**:The priority of Global.checkpoints is higher than the priority of Global.pretrain_weights, that is, when two parameters are specified at the same time, the model specified by Global.checkpoints will be loaded first. If the model path specified by Global.checkpoints is wrong, the one specified by Global.pretrain_weights will be loaded.
+
+
 ## Evaluation Indicator
 
 PaddleOCR calculates three indicators for evaluating performance of OCR detection task: Precision, Recall, and Hmean.
diff --git a/doc/doc_en/inference_en.md b/doc/doc_en/inference_en.md
index 3326bab87f83e5025f885e03204aeed769d04868..8932eafc64232584842a97571c5a7b0672a2ad55 100644
--- a/doc/doc_en/inference_en.md
+++ b/doc/doc_en/inference_en.md
@@ -182,6 +182,13 @@ self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
 dict_character = list(self.character_str)
 ```
 
+### 4.Recognition model inference using custom text dictionary file
+If the text dictionary is replaced during training, you need to specify the text dictionary path by setting the parameter `rec_char_dict_path` when using your inference model to predict.
+
+```
+python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./your inference model" --rec_image_shape="3, 32, 100" --rec_char_type="en" --rec_char_dict_path="your text dict path"
+```
+
 ## Text detection and recognition inference concatenation
 
 ### 1. Ultra-lightweight Chinese OCR model inference
diff --git a/ppocr/data/det/dataset_traversal.py b/ppocr/data/det/dataset_traversal.py
index ab635de3e9f22cd0556c651bf9c70654bd83be77..737cbe2e90dc259c1266b554ae0c6735ea3a52d2 100644
--- a/ppocr/data/det/dataset_traversal.py
+++ b/ppocr/data/det/dataset_traversal.py
@@ -95,8 +95,10 @@ class EvalTestReader(object):
             for img_path in img_list:
                 img = cv2.imread(img_path)
                 if img is None:
-                    logger.info("load image error:" + img_path)
+                    logger.info("{} does not exist!".format(img_path))
                     continue
+                elif len(list(img.shape)) == 2 or img.shape[2] == 1:
+                    img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
                 outs = process_function(img)
                 outs.append(img_path)
                 batch_outs.append(outs)
diff --git a/ppocr/data/det/db_process.py b/ppocr/data/det/db_process.py
index 0993324aa92a6765a324472f9137f266b2dabb52..ee131501724123ac2d718e596022c3ee266077a8 100644
--- a/ppocr/data/det/db_process.py
+++ b/ppocr/data/det/db_process.py
@@ -104,6 +104,8 @@ class DBProcessTrain(object):
         if imgvalue is None:
             logger.info("{} does not exist!".format(img_path))
             return None
+        if len(list(imgvalue.shape)) == 2 or imgvalue.shape[2] == 1:
+            imgvalue = cv2.cvtColor(imgvalue, cv2.COLOR_GRAY2BGR)
         data = self.make_data_dict(imgvalue, gt_label)
         data = AugmentData(data)
         data = RandomCropData(data, self.image_shape[1:])
diff --git a/ppocr/data/rec/img_tools.py b/ppocr/data/rec/img_tools.py
index 575432933bf5f261550636a8e1a24fba3ec33177..d6e273cba2db0d2be432841d5731aa11c0237c5d 100755
--- a/ppocr/data/rec/img_tools.py
+++ b/ppocr/data/rec/img_tools.py
@@ -15,6 +15,8 @@
 import math
 import cv2
 import numpy as np
+from ppocr.utils.utility import initial_logger
+logger = initial_logger()
 
 
 def get_bounding_box_rect(pos):
@@ -101,9 +103,13 @@ def process_image(img,
         norm_img = resize_norm_img_chinese(img, image_shape)
     norm_img = norm_img[np.newaxis, :]
     if label is not None:
-        char_num = char_ops.get_char_num()
+        # char_num = char_ops.get_char_num()
         text = char_ops.encode(label)
         if len(text) == 0 or len(text) > max_text_length:
+            logger.info(
+                "Warning in ppocr/data/rec/img_tools.py:line106: Wrong data type."
+                "Excepted string with length between 0 and {}, but "
+                "got '{}' ".format(max_text_length, label))
             return None
         else:
             if loss_type == "ctc":
diff --git a/tools/program.py b/tools/program.py
index 30e9d7379bd07a0570c613ea25d1b440ed8bc682..3c71065a167fa18fc9d00535dace97737904b74d 100755
--- a/tools/program.py
+++ b/tools/program.py
@@ -114,7 +114,10 @@ def merge_config(config):
                 global_config[key] = value
         else:
             sub_keys = key.split('.')
-            assert (sub_keys[0] in global_config), "the sub_keys can only be one of global_config: {}, but get: {}, please check your running command".format(global_config.keys(), sub_keys[0])
+            assert (
+                sub_keys[0] in global_config
+            ), "the sub_keys can only be one of global_config: {}, but get: {}, please check your running command".format(
+                global_config.keys(), sub_keys[0])
             cur = global_config[sub_keys[0]]
             for idx, sub_key in enumerate(sub_keys[1:]):
                 assert (sub_key in cur)
@@ -177,7 +180,6 @@ def build(config, main_prog, startup_prog, mode):
                 optimizer.minimize(opt_loss)
                 opt_loss_name = opt_loss.name
                 global_lr = optimizer._global_learning_rate()
-                global_lr.persistable = True
                 fetch_name_list.insert(0, "lr")
                 fetch_varname_list.insert(0, global_lr.name)
     return (dataloader, fetch_name_list, fetch_varname_list, opt_loss_name)