{"full_name":"PaddlePaddle/PaddleOCR","owner":"PaddlePaddle","name":"PaddleOCR","url":"https://github.com/PaddlePaddle/PaddleOCR","description":"Turn any PDF or image document into structured data for your AI. A powerful, lightweight OCR toolkit that bridges the gap between images/PDFs and LLMs. Supports 100+ languages.","homepage":"https://www.paddleocr.com","topics":[],"language":"Python","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/23534030?v=4","created_at":"2020-05-08T10:38:16Z","pushed_at":"2026-09-16T03:31:50Z","metrics":{"stars":92511,"forks":11640,"watchers":567,"issues":10755,"prs":4075,"releases":33,"issues_open":null,"issues_closed":null,"prs_open":null,"prs_closed":null,"prs_merged":null,"discussions":3421},"growth":{"day":{"stars":49,"forks":4,"base_day":"2026-11-09"}},"history":{"dates":["2026-11-09","2026-11-10"],"stars":[92462,92511],"forks":[11636,11640],"watchers":[567,567],"issues_open":[null,null],"prs_open":[null,null]},"boards":{"total":170},"same_owner":[{"full_name":"PaddlePaddle/Paddle-Lite","name":"Paddle-Lite","boards":{}},{"full_name":"PaddlePaddle/PaddleClas","name":"PaddleClas","boards":{}},{"full_name":"PaddlePaddle/PaddleSeg","name":"PaddleSeg","boards":{}},{"full_name":"PaddlePaddle/PaddleSpeech","name":"PaddleSpeech","boards":{}}],"issues":[{"number":16823,"title":"Frequently Asked Questions on Inference and Deployment of PaddleOCR-VL PaddleOCR-VL 推理部署相关高频问题回复","comments":56,"created_at":"2025-10-24T09:32:15Z"},{"number":17858,"title":"【Hackathon 10th】PaddleOCR 全球衍生模型挑战赛 / PaddleOCR Global Derivative Model Challenge","comments":45,"created_at":"2026-03-26T10:11:42Z"},{"number":16342,"title":"cudnn相关dll文件加载异常","comments":17,"created_at":"2025-08-27T03:18:02Z"},{"number":17640,"title":"使用 paddleocr 3.4.0，paddlepaddle-gpu 3.3.0，paddlex 3.4.1, PaddleOCR-VL-1.5对图片进行ocr识别,发现有时会出现长时间不返回识别结果问题(之前问题有复现,但是这边重新open不了对应issue)","comments":16,"created_at":"2026-02-05T01:06:19Z"},{"number":17539,"title":"Getting this error when picking a image","comments":14,"created_at":"2026-01-23T07:30:01Z"},{"number":18161,"title":"请问是放弃了HPI和TensorRT吗？","comments":13,"created_at":"2026-06-13T23:23:43Z"},{"number":18162,"title":"NotImplementedError when using PaddleOCR 3.7.0 with PaddlePaddle 3.3.1","comments":11,"created_at":"2026-06-14T04:00:13Z"},{"number":17190,"title":"记录一次在910B3 NPU上部署 PaddleOCR-VL的过程","comments":11,"created_at":"2025-11-26T09:45:45Z"}],"discussions":[{"number":18433,"title":"`return_word_box` on AI Studio","comments":0,"category":"Ideas & Features","created_at":"2026-11-05T23:12:33Z"},{"number":18425,"title":"Processing text detection output","comments":1,"category":"Q&A","created_at":"2026-10-29T12:34:14Z"},{"number":18424,"title":"Save as searchable PDF","comments":0,"category":"Ideas & Features","created_at":"2026-10-29T02:03:33Z"},{"number":18414,"title":"Does an app already exist that can automatically sort graduation photos by person?","comments":1,"category":"Q&A","created_at":"2026-10-25T21:21:45Z"},{"number":18405,"title":"License and compatibility clarification for PP-OCRv5 Japanese OCR model files","comments":0,"category":"Q&A","created_at":"2026-10-21T02:49:33Z"},{"number":18404,"title":"v3.0.0/version3.x/module_usage/text_image_unwarping","comments":1,"category":"Q&A","created_at":"2026-10-21T02:12:14Z"},{"number":18400,"title":"korean-handwriting-rec: korean_PP-OCRv5_mobile_rec fine-tuned on 4.6M Korean handwritten and printed words","comments":0,"category":"Show and tell","created_at":"2026-10-18T04:54:35Z"},{"number":18394,"title":"PPStructure V3 怎么支持并发部署","comments":2,"category":"Q&A","created_at":"2026-10-13T05:08:39Z"}],"latest_release":{"name":"v3.7.0","tag":"v3.7.0","published_at":"2026-06-11T12:09:14Z","body":"## 2026.6.11 v3.7.0 released\r\n\r\n* **Release PP-OCRv6**\r\n\r\n    - **Accuracy boost**: Medium tier achieves +4.6% detection and +5.1% recognition over PP-OCRv5_server, surpassing mainstream VLMs (Qwen3-VL-235B, GPT-5.5) with only 34.5M parameters.\r\n    - **50 languages unified**: Single model covers Chinese, English, Japanese, and 46 Latin-script languages — no model switching needed.\r\n    - **Specialized scenarios**: Major improvements in digital displays, dot-matrix characters, tire prints, and industrial text recognition.\r\n    - **Faster inference**: 5.2× CPU speedup (OpenVINO), 6.1× on Apple M4 (tiny), 0.13s on A100 GPU.\r\n    - **Three tiers for all scenarios**: tiny (1.5M) / small (7.7M) / medium (34.5M) for edge, mobile, and server deployment.\r\n\r\n## 2026.6.11 v3.7.0 发布\r\n\r\n* **发布 PP-OCRv6**\r\n\r\n    - **精度全面提升**：medium 档相比 PP-OCRv5_server 检测精度提升 4.6%、识别精度提升 5.1%，以仅 34.5M 参数超越 Qwen3-VL-235B、GPT-5.5 等主流视觉语言大模型。\r\n    - **50 种语言统一支持**：单一模型覆盖中文、英文、日文及 46 种拉丁语系语言，无需为不同语种切换模型。\r\n    - **专业场景增强**：数码显示屏、点阵字符、轮胎印字、工业字符等传统 VLM 难以覆盖的场景识别能力大幅提升。\r\n    - **推理速度更快**：medium 档 CPU OpenVINO 推理加速 5.2×，tiny 档 Apple M4 加速 6.1×，A100 上仅需 0.13s。\r\n    - **三档模型覆盖全场景**：tiny（1.5M）/ small（7.7M）/ medium（34.5M）分别面向端侧/移动端/服务端部署。\r\n\r\n**Full Changelog**: https://github.com/PaddlePaddle/PaddleOCR/compare/v3.6.0...v3.7.0"},"readme":true,"readme_zh":false,"news":[{"title":"PaddleOCR, Multilingual OCR Tools","url":"https://github.com/PaddlePaddle/PaddleOCR","hn":"https://news.ycombinator.com/item?id=25622609","source":"Hacker News","lang":"en","date":"2021-01-03","points":19,"comments":3},{"title":"PaddleOCR: Multilingual OCR toolkits based on PaddlePaddle","url":"https://github.com/PaddlePaddle/PaddleOCR","hn":"https://news.ycombinator.com/item?id=40623307","source":"Hacker News","lang":"en","date":"2024-06-09","points":9,"comments":2},{"title":"PaddleOCR 업데이트: 이미지 및 PDF를 LLM 대응 데이터로 변환하는 차세대 OCR 도구","url":"https://note.com/humble_bobcat51/n/n4ce7e23ea83a?hl=ko","source":"note","lang":"en","date":"2026-05-29"},{"title":"PaddleOCR Becomes Most-Starred OCR Project on GitHub, Surpassing Google’s Tesseract","url":"https://pandaily.com/paddle-ocr-becomes-most-starred-ocr-project-on-git-hub-surpassing-google-s-tesseract","source":"pandaily","lang":"en","date":"2026-03-29"},{"title":"Training model from AIDA-project","url":"https://huggingface.co/Kansallisarkisto/PaddleOCR_training","source":"huggingface","lang":"en","date":"2024-08-27"},{"title":"百度文心发布 PaddleOCR-VL-1.6：准确率突破 96.33%，刷新文档解析 SOTA","url":"https://news.google.com/rss/articles/CBMiVEFVX3lxTFBZTUpnX0dZeXdNdm1GSFhueDBRNVhTbU9hM25IOGl5X3RGbV9hbThGNS11UDhkS0lVRE14VnhNWW84RDl5NTNCRHlJcHlrdlE0UHJWNQ?oc=5","source":"量子位","lang":"zh","date":"2026-06-02"},{"title":"文心衍生模型PaddleOCR创GitHub“星”数记录","url":"https://news.google.com/rss/articles/CBMiZEFVX3lxTE1xV05uT3VqeThVdUo4Q2hrVmg1bmtHVHFmOTZscjd4SzUzY0FhNXg0b1hVa0pobmFadTk1bjh6ZWJtX1Q0ZVpRamdIZjh0RFRxaVVZS0V3WC04QVFVb2JFVEotMjc?oc=5","source":"科学网—新闻","lang":"zh","date":"2026-03-31"},{"title":"百度击败谷歌！PaddleOCR成GitHub最受欢迎OCR项目：中国开源首度登顶","url":"https://news.google.com/rss/articles/CBMiWEFVX3lxTE5RVjNQYkdodUc2MDVGX3ZhWG8yZU1zQWsyZl9NSGhfU3IzU1BZUzgxN3BfRnN3TXMyNnR1dlo0Y2xPRUVabzQ1cW4xckE2NDR1U1lpdU5fdnA?oc=5","source":"驱动之家","lang":"zh","date":"2026-03-31"},{"title":"PaddleOCR Tops GitHub Star Global First: Chinese Open Source Power Leads in OCR Field","url":"https://news.google.com/rss/articles/CBMiSkFVX3lxTE1jazBFczZESkxZOE1LSnhfbGkzZ1VlbHpMZHJYUWhtU2RURmhQbUNocEF3YkJDOWV6cjJ3UEtWdWVLdjl5LXpPaTh3?oc=5","source":"AIBase","lang":"en","date":"2026-03-30"},{"title":"超越谷歌 中国开源项目PaddleOCR登顶GitHub最受欢迎OCR项目","url":"https://news.google.com/rss/articles/CBMicEFVX3lxTE85NUF4ODFRZm94cTVfUHpad05CMEstcVphdkI4YnVRTnhEeGpSX2s4Z1NwMk8tSFJRZ2ExaUlsZ2xOcC1uaTc4amV5R29SWlAybEJCMTctTUNFV1RreTFqM1J3aVNEM0tNbmFscVFzaUk?oc=5","source":"中国科技网","lang":"zh","date":"2026-03-30"},{"title":"超越谷歌Tesseract OCR！PaddleOCR登顶GitHub Star全球第一","url":"https://news.google.com/rss/articles/CBMiX0FVX3lxTE8zSDVIWEdvSHdxajhHM2U3djE0U2F0VTdSdmtkQlgyZXB6ZzF5VGtVTGI2OW1aZ2ttYWtNMXNDaUVvX1Y0dU1ZUWxTcG0zM0JMdUZta2RVTzhvazdKNUdz?oc=5","source":"OSCHINA","lang":"zh","date":"2026-03-30"},{"title":"5 Open Source Tools You Can Use to Train and Deploy an OCR Project","url":"https://news.google.com/rss/articles/CBMisAFBVV95cUxNc0c4Tk1xUFJhSVVoMGNwelppT3A2SzM1YXFBUjgtZDlXVnFPOWxmS2o3cmtyTDhsSmtwOHBmSkFicjZIMnJMWXJyVFJqYlNMdGhfX1U4YWZ2S0ZQeW5CajN2bU5UbkQwNTVfaXZhQTh3N0FVN085ZlRWUVpaUlFMdUM5bGRzR2wwYVNwcEFWdnZCMmRQekJwR2pVTVUtNkFRWFRhVTZvR1RCRzhLWW5NTg?oc=5","source":"Towards Data Science","lang":"en","date":"2025-02-04"},{"title":"OCR Fine-Tuning: From Raw Data to Custom Paddle OCR Model","url":"https://news.google.com/rss/articles/CBMihgFBVV95cUxOdExfUW9zZy1QZm81STVFbmdhQWNJSmVaSDl3MXlEMktlV00wM2ZwdlhmU0VjRVIzczVtOWx1SS1tNnBJTDJpV2hpUWZCMWJ5Y3ZrcmJmU2RvUlZQRk5rZEZuaGV4ZFNTLXBPZXJvYkJMbGlDVHVkUmszY3ZoRV85VkZPeEh5UQ?oc=5","source":"HackerNoon","lang":"en","date":"2025-01-28"},{"title":"Onboard PaddleOCR with Amazon SageMaker Projects for MLOps to perform optical character recognition on identity documents","url":"https://news.google.com/rss/articles/CBMi_AFBVV95cUxPZkw3dVhSa3puaVFnbF9wbFBRX3I1cGgwUUhPUTBnblVmZm4zamFnbFVRSE5qU3hPbm9DNDFBME5tTEtfek1PckYwQU9qUjZIX0FiNU5hRkN1bnY0Vi0zdGFkenFhYlc3UEtNSU84azdyeFUydjVqeFR0dHNzUUJCQUQyZE1MYlVWRFVqRkhmRWVLcF91TjBIY3JOdnBkTmdrNklDYVNkdG16VzhsamxacHRGTDZNbW9MTUc1RzU3aDZFNWpSZzdmVnJVdFZYTW8xaEw1Nk9sZTkxdnpfRHVfQ0QtZGE4bGxJeXZQNExESnF4S3VzUGdHVjIzZGg?oc=5","source":"Amazon Web Services (AWS)","lang":"en","date":"2022-07-08"}],"stale":false,"snapshot_at":"2026-11-10T09:35:25Z","history_start":"2026-09-06"}