mtfyun/rapidocr
1
0 Fork 0
mirror of https://github.com/RapidAI/RapidOCR.git synced 2026-11-11 21:40:52 +08:00
Code Issues Packages Projects Releases Wiki Activity
1,403 Commits 6 Branches 70 Tags
20cbc41e36a135ff972c5e40e4637badb2c770ce
Commit Graph
136 Commits
Author SHA1 Message Date
SWHL 1518e72235 chore: update files 2025-01-20 21:18:56 +08:00
SWHL 8b81b148b1 chore: update files 2025-01-20 20:54:29 +08:00
SWHL c2b409f023 chore: update files 2025-01-19 22:01:25 +08:00
SWHL 94576e5e4c chore: optimize code 2025-01-17 09:36:38 +08:00
Joker1212 f699fdd3a9 feat: support det mean and std config (#312) 2025-01-11 17:11:28 +08:00
SWHL f53c64007c feat(rapidocr_api): add uvicorn timestamp 2025-01-01 16:41:30 +08:00
weijh 8913c7ac1b feat(rapidocr_api): 为uvicorn日志添加时间戳 (#309) 
* Update README.md

* uvicorn 日志加时间戳

* Update README.md

* add vim for edit files in container

* Create readme.md

* Update readme.md

* Update README.md

* Delete python/rapidocr_paddle/ch_ppocr_det/readme.md

* Create readme.md

* Update readme.md

* Update readme.md
 2025-01-01 16:06:00 +08:00
SWHL 7f5594bbca fix(rapidocr_onnxruntime): merged PR #294 2024-12-12 21:32:11 +08:00
jesse01andjesse01 0d70e86a08 feat(rapidocr_onnxruntime): 优化单字符坐标及返回单字符置信度 (#293) 
* 1、utils.py：优化单字符坐标计算。根据CTC解码特点，增加每个字符长度列表作为判断条件，一般非中文字符2-3个解码宽度，中文3-4个解码宽度，当某个字符与上一字符距离大于4个解码宽度，表示中间有空白，此时进行字段拆分，这样可以减少文字间空白对坐标计算的影响；同时返回每个字符值信度，以方便应用使用

2、cal_rec_boxes/main.py：
 配合返回单字符值信度列表的修改
 修改计算单字符宽度的分母为总字符数减1，与分母一致，这样计算出的值更加准确

 3.main.py：
 修复主程序没有传递命令行参数的问题

* 优化文本行检测优化，当limit_type为max，根据图片大小动态设置limit_side_len，以减少漏检

---------

Co-authored-by: jesse01 <jesse01.163.com>
 2024-12-12 09:31:58 +08:00
SWHL 3da032a00e fix(rapidocr_onnxruntime): fixed issue #288 2024-12-09 09:28:45 +08:00
SWHL cababcf25c feat(rapidocr_openvino): support en char rec 2024-12-03 22:15:25 +08:00
Joker1212 01a5a65952 feat(rapidocr_onnxruntime): support en char rec (#272) 
* feat: sup en char rec

* test: add en char rec test
 2024-12-03 09:30:54 +08:00
SWHL df36c69820 chore: optimze char rec (#260) 
* chore: format code

* chore: format code

* test: update unit tests for different platform

* feat(rapidocr_onnxruntime): optimize code

* feat(rapidocr_onnxruntime): add --return_word_box params of CLI

* chore: update demo

* feat(rapidocr_onnxruntime): update vis of return_word_box
 2024-11-19 22:27:18 +08:00
xinyi 2ac3c35a36 feat: char rec(单字识别) (#254) 
* feat: add word spilit for ocr

* chore: rm unnecessary change

* test: fix test error
 2024-11-13 09:23:22 +08:00
SWHL 2be4feff82 fix(python): fix incorrect usage of the exists function 2024-11-13 08:25:06 +08:00
Pumpkinor f7978c650b chore(rapidocr_onnxruntime): change read_yaml order (#251) 
fisrt check the input args 
if the param named 'config_path'  exists, use 'config_path' first.
if it is not exists, use the default one named 'DEFAULT_CFG_PATH'
 2024-11-13 08:21:20 +08:00
SWHL ce06654ac0 fix(rapidocr_onnxruntime): fix normalize value of det (#246) 2024-10-31 09:17:57 +08:00
SWHL c3504c2f7d chore: Compatible with python 3.7 2024-10-17 22:51:03 +08:00
SWHL 5e1b5ef196 chore: Compatible with python 3.7 2024-10-17 22:47:19 +08:00
SWHL fd520ffdad Fix issue 232 (#235) 
* fix(rapidocr_onnxruntime): Fixed padding_h variable before assignment

* feat(rapidocr_onnxruntime): Add resize function

* feat(rapidocr_onnxruntime): Add limit of max_side_len and min_side_len

* tests: Add __init__
 2024-10-17 22:24:31 +08:00
SWHL 6b60256ee7 chore: Optimize code structure 2024-09-06 10:13:12 +08:00
SWHL 1de4db0249 refactor(rapidocr_onnxruntime): Refactor the pre-processing code for the det part 2024-06-27 10:02:17 +08:00
SWHL ac5547345b refactor(rapidocr_onnxruntime): Add typing annotations of RapidOCR call 2024-06-09 17:01:58 +08:00
SWHL e7094936a9 refactor(rapidocr_onnxruntime): Add typing annotations and remove invalid code 2024-06-09 16:32:41 +08:00
SWHL 35e85ef370 fix(rapidocr_onnxruntime): Fix EP_list parameter error 2024-05-21 18:21:50 +08:00
SkeathyTomas 90b799c6db fix(rapidocr_onnxruntime): fixed EP error (#183) 2024-05-21 13:59:59 +08:00
SWHL f98585952b fix(python): Add the default value of maxsize in lru_cache under python3.7 for compatibility 2024-05-21 08:18:30 +08:00
SWHL cb14cc846b style(rapidocr_onnxruntime): Code wrapping with black 2024-05-21 08:06:25 +08:00
SWHL a291ca1dcb chore(rapidocr_onnxruntime): Optim tips 2024-05-20 21:46:01 +08:00
SWHL 5891ba2142 chore(rapidocr_onnxruntime): Add logger cache to reduce print nums 2024-05-20 21:45:04 +08:00
SkeathyTomas fb23f82870 fix(rapidocr_onnxruntime): Fix dml not work issue (#180) 2024-05-20 21:06:39 +08:00
Dolen b9c90ee837 fix(python): Modify the way CTCLabelDecode calculates confidence (#179) 2024-05-18 15:33:20 +08:00
SWHL 7528fa9660 chore(rapidocr_onnxruntime): Optim tips of infer_engine 2024-05-18 15:27:13 +08:00
SWHL de44841377 refactor(rapidocr_onnxruntime): Decoupling utils.py files 2024-05-18 15:12:10 +08:00
SWHL d60fe5b415 chore(rapidocr_onnxruntime): Optim tips about OrtInferSession 2024-05-17 16:34:16 +08:00
SWHL 0e1cb1814f chore(rapidocr_onnxruntime): Add friendly tips for using cuda 2024-05-16 22:17:21 +08:00
SWHL 91e40ad4c3 fix(rapidocr_onnxruntime): Optimize prompt when using DirctML 2024-05-16 22:12:42 +08:00
SWHL 7a731483a5 fix(config): Remove the default installation onnxruntime-directml restriction on Windows. 2024-05-15 22:24:38 +08:00
Ning hao 8523a2e453 fix(config): Make ONNX Runtime DirectML an optional configuration for better compatibility (#178) 
This commit introduces an optional configuration for using ONNX Runtime DirectML.

DirectML can significantly improve performance on newer devices, but it has stricter requirements and may not be suitable for all users.

By making it an optional configuration, we ensure better compatibility for users with older devices or who are running the CLI version, where the GPU's potential may not be fully utilized due to single inference models.

For more information on DirectML requirements, please refer to: https://github.com/microsoft/DirectML
 2024-05-15 21:43:03 +08:00
SWHL c5c2246ed1 Optim OrtInferSession 2024-05-12 14:50:19 +08:00
Ning hao 01430ed41e feat(cli): Support GPU acceleration via ONNX Runtime DirectML on Windows (#176) 
* feat(cli): Support GPU acceleration via ONNX Runtime DirectML on Windows

This commit adds support for GPU acceleration in the CLI using ONNX Runtime DirectML. This allows users with compatible GPUs on Windows to leverage their hardware for faster inference.

* fix small bug
 2024-05-12 10:53:19 +08:00
SWHL e4b66000b0 Add unit testing about fixed issue #170 2024-04-19 09:59:49 +08:00
HAL9000COM f602d301e5 Bug fix for converting error when processing bitmap image. (#171) 2024-04-19 09:48:21 +08:00
SWHL 182e54ef6d Fixed issue #161 2024-04-07 09:51:57 +08:00
SWHL f3732c102e Add support for RapidOCR class init to specify the number of threads. 2024-03-07 09:41:17 +08:00
SWHL 55bb974b5d Support transparent white font image rec in rapidocr_onnxruntime 2024-03-06 10:40:17 +08:00
SWHL bd2eba459f Update use_cuda parameter in config.yaml 2024-03-04 22:59:40 +08:00
SWHL 2f66907a6c Add intra_op_num_threads and inter_op_num_threads in config.yaml 2024-03-04 22:52:47 +08:00
SWHL 2caa0b083f Update LoadImage Class and support Image as input. 2024-02-28 18:21:02 +08:00
SWHL e55c3a2a37 Adapt to Pillow v9 and v10 2024-02-27 15:11:01 +08:00