📣 Subsequent update logs will be moved to Release, and will no longer be updated here

🚩2023-10-08 v2.2.8 update

  • Adapt the relevant parameters of rapidocr_onnxruntime, which can be passed in through the RapidVideOCR class, so as to more flexibly specify models of different languages.

♦ 2023-08-05 v2.2.4 update

  • Fix the index error in batch recognition mode.

  • Add a logging module to facilitate the use of the desktop version, quickly record problems, and facilitate feedback.

🛶2023-07-19 v2.2.3 update

  • Added adaptation to VSF parameters. When in command line mode and during class initialization, you can specify the same-name parameters of the VSF command. For detailed usage, please refer to link

🤓2023-07-08 v2.2.2 update

  • Fixed the problem that Chinese paths cannot be read during batch recognition
  • Fixed the problem of skipping in SRT when missing axes. Currently, when an axis fails to be recognized, a position will be vacated for easy proofreading.
  • Keep the intermediate results of VSF recognition

🐲2023-06-22 v2.2.0 update

  • This version is compatible with v2.1.x, which means that the previous usage is still possible.
  • Integrate VSF’s CLI into the library, just specify the full path of VideoSubFinderWXW.exe.
  • Added batch recognition function, specify the video directory, and automatically extract all video subtitles in the directory
  • For usage examples, see: demo.py

😀2023-05-12 v2.1.7 update

  • Optimized code
  • Added save_name parameter, which can flexibly specify the name of the saved srt | txt file, the default is result

🐱2023-03-27 v2.1.6 update

  • Fixed the timeline misalignment problem, see issue 23 for details

👽2023-03-23 v2.1.5 update

  • Added control parameter is_print_console for printing to the screen
  • Adjust the out_format parameter position to the initialization class

😀2023-03-14 v2.1.3 update

  • Fix the error when passing in the TXTImages directory and identifying duplicate characters

😜2023-03-12 v2.1.2 update

  • Fix index error, #22

🎢2023-03-11 v2.1.1 update

  • Fix the difference between single image recognition and previous versions

  • The default recognition mode is changed to single image recognition. Whether to use duplicate image recognition is up to you

🥇2023-03-10 v2.1.0 update

  • Added duplicate character recognition function, faster, and the default is duplicate character recognition function

🎈2023-03-02 v2.0.5~7 update

  • Fix format errors in generated srt files, #19

🎫2023-02-17 v2.0.4 update

  • Optimized passing in the TXTImages directory. Compared with passing in RGBImages, it will be faster and more accurate. It is recommended to pass in the TXTImages directory

💎2023-02-17 v2.0.2 update

  • Fix the problem of missing spaces in peer subtitle recognition

🎈2023-01-29 v1.1.10 update

  • Fix the error of empty index when converting frame index to timestamp

🧨2023-01-28 v1.1.9 update

  • Fix the problem of timeline misalignment

👊 2023-01-15 v1.1.4 update

  • Add the option of outputting txt format. Currently, the v1.1.4 version outputs srt and txt formats by default

  • Add the option to adjust the box size of subtitle selection according to the screen size of the running program

🌈2023-01-10 v1.0.3 update

  • Replace decord with OpenCV because decord has a memory leak when processing MP4. For details, see: #208

🎄2022-12-04 update

  • Added interactive subtitle positioning function, which is enabled by default and is more user-friendly. For details, please refer to the GIF below. Thanks to @Johndirr for the suggestion.
  • Optimize the code structure, put RapidOCR related models and configuration files in the rapidocr directory
  • The configuration files of rapidvideocr are also placed in the corresponding directory

🌼2022-05-08 update

  • Add interactive determination of the threshold of the binary subtitle image, which only supports Windows and can be used by is_select_threshold = True
  • Optimize the code

🎉2022-05-03 update

💡2022-05-01 update

  • Added speech module part is located in the branch asr_module

  • Added speech recognition module. Since the decoding part of this module can only run on Linux and Mac, if you want to use this module, please use Linux and Mac.

  • Currently, the speech recognition code comes from the RapidASR/python part. The model comes from PaddleSpeech

  • After a simple test, the speech recognition module is not very accurate. -_-!

2022-03-09 update

  • Added FAQ module to help everyone skip common small problems

2021-12-14 update

  • Background removal is not effective, resulting in the loss of some frames

  • Tried to use the image segmentation method. After testing, the inference speed under CPU is too slow, so it is temporarily abandoned

  • Currently using a fixed binary threshold

  • (2021-12-14) Improve the corresponding English document

  • (2021-12-14) Add running time benchmark

  • Add specific parameter description

  • Make a project logo

  • More tests

Edit this page

Last updated 21 May 2025, 19:10 -0600 . history