YaoFANGUK /video-subtitle-remover

About video-subtitle-remover

VSR is a local app that erases hard-coded subtitles and text watermarks from video and images. It finds the text with OCR, then fills those spots with AI inpainting instead of blurring. GUI and CLI both ship, and it runs offline with no paid APIs. Most of the docs are in Chinese, and ProPainter mode eats a lot of video memory.

Best for: Remix editors who need clean footage from subtitled or watermarked clips before re-editing or dubbing.

• Removes hard-coded subtitles and text watermarks at the original resolution
• Fills removed areas with AI inpainting, not mosaic or nearby pixels
• Lets you pass subtitle coordinates to remove text in one area only
• Removes all text in the frame when you don't pass coordinates
• Cleans watermarks from many images in one batch
• Extracts the original subtitles, so you can pair it with VSE

1. 1
   Download a prebuilt package from Releases and unpack it, or install Python 3.12+ for a source setup
2. 2
   Create and activate a virtual environment: python -m venv videoEnv, then source videoEnv/bin/activate (Windows: videoEnv\Scripts\activate)
3. 3
   Install the packages for your hardware, e.g. CUDA: the command below, the next command, then the next command

   pip install paddlepaddle-gpu==3.0.0 -i https://www.paddlepaddle.org.cn/packages/stable/cu118/

   Copy

   pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu118

   Copy

   pip install -r requirements.txt

   Copy
4. 4
   Start the GUI with python gui.py, or the CLI with python ./backend/main.py
5. 5
   Or use Docker: the command below

   docker run -it --name vsr --gpus all eritpchy/video-subtitle-remover:1.4.0-cuda11.8 python backend/main.py -i test/test.mp4 -o test/test_no_sub.mp4

   Copy