Skip to content

Latest commit

 

History

History
21 lines (13 loc) · 1010 Bytes

README.md

File metadata and controls

21 lines (13 loc) · 1010 Bytes

Convert hardcoded subtitles from videos to text files

Warning: this project is still under development. It is only suitable for 1080p videos, and the coodinates of subtitles on video frames is hardcoded as well as the subtitles themselves.

Usage

  1. Install the dependencies using pip:

    pip install progressbar2 baidu-aip opencv-python-headless numpy

  2. Edit config.json file.

    • Note to change the value of videosDir to the path of videos folder on your computer.
    • splitDuration is the time intervals between every processed frame from all frames in a video. The default value is 1 second.
    • Please create an app using OCR APIs from Baidu AI Platform, and fill in the AppID, API Key and Secret Key in config.json.

  3. Run python get_frames.py to get subtitle images from videosDir and save them to a folder(subtitleImgsDir in config.json)

  4. Run python ocr_subtitles.py <video_name> to convert subtitle images to a text file in resultsDir.