diff --git a/applications/README.md b/applications/README.md index 017c2a9f6f696904e9bf2f1180104e66c90ee712..2637cd6eaf0c3c59d56673c5e2d294ee7fca2b8b 100644 --- a/applications/README.md +++ b/applications/README.md @@ -20,10 +20,10 @@ PaddleOCR场景应用覆盖通用，制造、金融、交通行业的主要OCR ### 通用 -| 类别 | 亮点 | 模型下载 | 教程 | -| ---------------------- | ------------ | -------------- | --------------------------------------- | -| 高精度中文识别模型SVTR | 比PP-OCRv3识别模型精度高3%，可用于数据挖掘或对预测效率要求不高的场景。| [模型下载](#2) | [中文](./高精度中文识别模型.md)/English | -| 手写体识别 | 新增字形支持 | | | +| 类别 | 亮点 | 模型下载 | 教程 | 示例图 | +| ---------------------- | ------------------------------------------------------------ | -------------- | --------------------------------------- | ------------------------------------------------------------ | +| 高精度中文识别模型SVTR | 比PP-OCRv3识别模型精度高3%，
可用于数据挖掘或对预测效率要求不高的场景。 | [模型下载](#2) | [中文](./高精度中文识别模型.md)/English | | +| 手写体识别 | 新增字形支持 | [模型下载](#2) | [中文](./手写文字识别.md)/English | | @@ -42,14 +42,14 @@ PaddleOCR场景应用覆盖通用，制造、金融、交通行业的主要OCR ### 金融 -| 类别 | 亮点 | 模型下载 | 教程 | 示例图 | -| -------------- | ------------------------ | -------------- | ----------------------------------- | ------------------------------------------------------------ | -| 表单VQA | 多模态通用表单结构化提取 | [模型下载](#2) | [中文](./多模态表单识别.md)/English | | -| 增值税发票 | 尽请期待 | | | | -| 印章检测与识别 | 端到端弯曲文本识别 | | | | -| 通用卡证识别 | 通用结构化提取 | | | | -| 身份证识别 | 结构化提取、图像阴影 | | | | -| 合同比对 | 密集文本检测、NLP串联 | | | | +| 类别 | 亮点 | 模型下载 | 教程 | 示例图 | +| -------------- | ----------------------------- | -------------- | ------------------------------------- | ------------------------------------------------------------ | +| 表单VQA | 多模态通用表单结构化提取 | [模型下载](#2) | [中文](./多模态表单识别.md)/English | | +| 增值税发票 | 关键信息抽取，SER、RE任务训练 | [模型下载](#2) | [中文](./发票关键信息抽取.md)/English | | +| 印章检测与识别 | 端到端弯曲文本识别 | | | | +| 通用卡证识别 | 通用结构化提取 | | | | +| 身份证识别 | 结构化提取、图像阴影 | | | | +| 合同比对 | 密集文本检测、NLP串联 | | | | diff --git a/applications/README_en.md b/applications/README_en.md new file mode 100644 index 0000000000000000000000000000000000000000..95c56a1f740faa95e1fe3adeaeb90bfe902f8ed8 --- /dev/null +++ b/applications/README_en.md @@ -0,0 +1,79 @@ +English| [简体中文](README.md) + +# Application + +PaddleOCR scene application covers general, manufacturing, finance, transportation industry of the main OCR vertical applications, on the basis of the general capabilities of PP-OCR, PP-Structure, in the form of notebook to show the use of scene data fine-tuning, model optimization methods, data augmentation and other content, for developers to quickly land OCR applications to provide demonstration and inspiration. + +- [Tutorial](#1) + - [General](#11) + - [Manufacturing](#12) + - [Finance](#13) + - [Transportation](#14) + +- [Model Download](#2) + + + +## Tutorial + + + +### General + +| Case | Feature | Model Download | Tutorial | Example | +| ---------------------------------------------- | ---------------- | -------------------- | --------------------------------------- | ------------------------------------------------------------ | +| High-precision Chineses recognition model SVTR | New model | [Model Download](#2) | [中文](./高精度中文识别模型.md)/English | | +| Chinese handwriting recognition | New font support | [Model Download](#2) | [中文](./手写文字识别.md)/English | | + + + +### Manufacturing + +| Case | Feature | Model Download | Tutorial | Example | +| ------------------------------ | ------------------------------------------------------------ | -------------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | +| Digital tube | Digital tube data sythesis, recognition model fine-tuning | [Model Download](#2) | [中文](./光功率计数码管字符识别/光功率计数码管字符识别.md)/English | | +| LCD screen | Detection model distillation, serving deployment | [Model Download](#2) | [中文](./液晶屏读数识别.md)/English | | +| Packaging production data | Dot matrix character synthesis, overexposure and overdark text recognition | [Model Download](#2) | [中文](./包装生产日期识别.md)/English | | +| PCB text recognition | Small size text detection and recognition | [Model Download](#2) | [中文](./PCB字符识别/PCB字符识别.md)/English | | +| Meter text recognition | High-resolution image detection fine-tuning | [Model Download](#2) | | | +| LCD character defect detection | Non-text character recognition | | | | + + + +### Finance + +| Case | Feature | Model Download | Tutorial | Example | +| ----------------------------------- | -------------------------------------------------- | -------------------- | ------------------------------------- | ------------------------------------------------------------ | +| Form visual question and answer | Multimodal general form structured extraction | [Model Download](#2) | [中文](./多模态表单识别.md)/English | | +| VAT invoice | Key information extraction, SER, RE task fine-tune | [Model Download](#2) | [中文](./发票关键信息抽取.md)/English | | +| Seal detection and recognition | End-to-end curved text recognition | | | | +| Universal card recognition | Universal structured extraction | | | | +| ID card recognition | Structured extraction, image shading | | | | +| Contract key information extraction | Dense text detection, NLP concatenation | | | | + + + +### Transportation + +| Case | Feature | Model Download | Tutorial | Example | +| ----------------------------------------------- | ------------------------------------------------------------ | -------------------- | ----------------------------------- | ------------------------------------------------------------ | +| License plate recognition | Multi-angle images, lightweight models, edge-side deployment | [Model Download](#2) | [中文](./轻量级车牌识别.md)/English | | +| Driver's license/driving license identification | coming soon | | | | +| Express text recognition | coming soon | | | | + + + +## Model Download + +- For international developers: We're building a way to download these trained models, and since the current tutorials are Chinese, if you are good at both Chinese and English, or willing to polish English documents, please let us know in [discussion](https://github.com/PaddlePaddle/PaddleOCR/discussions). +- For Chinese developer: If you want to download the trained application model in the above scenarios, scan the QR code below with your WeChat, follow the PaddlePaddle official account to fill in the questionnaire, and join the PaddleOCR official group to get the 20G OCR learning materials (including "Dive into OCR" e-book, course video, application models and other materials) + +
+ +
+ + If you are an enterprise developer and have not found a suitable solution in the above scenarios, you can fill in the [OCR Application Cooperation Survey Questionnaire](https://paddle.wjx.cn/vj/QwF7GKw.aspx) to carry out different levels of cooperation with the official team **for free**, including but not limited to problem abstraction, technical solution determination, project Q&A, joint research and development, etc. If you have already used paddleOCR in your project, you can also fill out this questionnaire to jointly promote with the PaddlePaddle and enhance the technical publicity of enterprises. Looking forward to your submission! + + +trackgit-views + diff --git "a/applications/\345\217\221\347\245\250\345\205\263\351\224\256\344\277\241\346\201\257\346\212\275\345\217\226.md" "b/applications/\345\217\221\347\245\250\345\205\263\351\224\256\344\277\241\346\201\257\346\212\275\345\217\226.md" index cd7fa1a0b3c988b21b33fe8f123e7d7c3e851ca5..14a6a1c8f1dd2350767afa162063b06791e79dd4 100644 --- "a/applications/\345\217\221\347\245\250\345\205\263\351\224\256\344\277\241\346\201\257\346\212\275\345\217\226.md" +++ "b/applications/\345\217\221\347\245\250\345\205\263\351\224\256\344\277\241\346\201\257\346\212\275\345\217\226.md" @@ -279,6 +279,12 @@ LayoutXLM与VI-LayoutXLM针对该场景的训练结果如下所示。 可以看出，对于VI-LayoutXLM相比LayoutXLM的Hmean高了1.3%。 +如需获取已训练模型，请扫码填写问卷，加入PaddleOCR官方交流群获取全部OCR垂类模型下载链接、《动手学OCR》电子书等全套OCR学习资料🎁 + +
+ +
+ #### 4.4.3 模型评估