Introduction to PP-OCRv5¶
PP-OCRv5 is the new generation text recognition solution of PP-OCR, focusing on multi-scenario and multi-text type recognition. In terms of text types, PP-OCRv5 supports 5 major mainstream text types: Simplified Chinese, Chinese Pinyin, Traditional Chinese, English, and Japanese. For scenarios, PP-OCRv5 has upgraded recognition capabilities for challenging scenarios such as complex Chinese and English handwriting, vertical text, and uncommon characters. On internal complex evaluation sets across multiple scenarios, PP-OCRv5 achieved a 13 percentage point end-to-end improvement over PP-OCRv4.
Key Metrics¶
1. Text Detection Metrics¶
| Model | Handwritten Chinese | Handwritten English | Printed Chinese | Printed English | Traditional Chinese | Ancient Text | Japanese | General Scenario | Pinyin | Rotation | Distortion | Artistic Text | Average |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PP-OCRv5_server_det | 0.803 | 0.841 | 0.945 | 0.917 | 0.815 | 0.676 | 0.772 | 0.797 | 0.671 | 0.8 | 0.876 | 0.673 | 0.827 |
| PP-OCRv4_server_det | 0.706 | 0.249 | 0.888 | 0.690 | 0.759 | 0.473 | 0.685 | 0.715 | 0.542 | 0.366 | 0.775 | 0.583 | 0.662 |
| PP-OCRv5_mobile_det | 0.744 | 0.777 | 0.905 | 0.910 | 0.823 | 0.581 | 0.727 | 0.721 | 0.575 | 0.647 | 0.827 | 0.525 | 0.770 |
| PP-OCRv4_mobile_det | 0.583 | 0.369 | 0.872 | 0.773 | 0.663 | 0.231 | 0.634 | 0.710 | 0.430 | 0.299 | 0.715 | 0.549 | 0.624 |
Compared to PP-OCRv4, PP-OCRv5 shows significant improvement in all detection scenarios, especially in handwriting, ancient texts, and Japanese detection capabilities.
2. Text Recognition Metrics¶
| Evaluation Set Category | Handwritten Chinese | Handwritten English | Printed Chinese | Printed English | Traditional Chinese | Ancient Text | Japanese | Confusable Characters | General Scenario | Pinyin | Vertical Text | Artistic Text | Weighted Average |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PP-OCRv5_server_rec | 0.5807 | 0.5806 | 0.9013 | 0.8679 | 0.7472 | 0.6039 | 0.7372 | 0.5946 | 0.8384 | 0.7435 | 0.9314 | 0.6397 | 0.8401 |
| PP-OCRv4_server_rec | 0.3626 | 0.2661 | 0.8486 | 0.6677 | 0.4097 | 0.3080 | 0.4623 | 0.5028 | 0.8362 | 0.2694 | 0.5455 | 0.5892 | 0.5735 |
| PP-OCRv5_mobile_rec | 0.4166 | 0.4944 | 0.8605 | 0.8753 | 0.7199 | 0.5786 | 0.7577 | 0.5570 | 0.7703 | 0.7248 | 0.8089 | 0.5398 | 0.8015 |
| PP-OCRv4_mobile_rec | 0.2980 | 0.2550 | 0.8398 | 0.6598 | 0.3218 | 0.2593 | 0.4724 | 0.4599 | 0.8106 | 0.2593 | 0.5924 | 0.5555 | 0.5301 |
A single model can cover multiple languages and text types, with recognition accuracy significantly ahead of previous generation products and mainstream open-source solutions.
PP-OCRv5 Demo Examples¶
Deployment and Secondary Development¶
- Multiple System Support: Compatible with mainstream operating systems including Windows, Linux, and Mac.
- Multiple Hardware Support: Besides NVIDIA GPUs, it also supports inference and deployment on Intel CPU, Kunlun chips, Ascend, and other new hardware.
- High-Performance Inference Plugin: Recommended to combine with high-performance inference plugins to further improve inference speed. See High-Performance Inference Guide for details.
- Service Deployment: Supports highly stable service deployment solutions. See Service Deployment Guide for details.
- Secondary Development Capability: Supports custom dataset training, dictionary extension, and model fine-tuning. Example: To add Korean recognition, you can extend the dictionary and fine-tune the model, seamlessly integrating into existing production lines. See Text Recognition Module Usage Tutorial for details.