video-subtitle-remover
An open-source local AI tool for removing hard-coded subtitles and watermarks from videos. Written in Python, it uses image inpainting algorithms to erase subtitles without resolution loss and reconstructs the background. It requires no third-party APIs and supports both image and video processing.
Timeline (last 90 days)
video-subtitle-remover (VSR) is an open-source project that uses AI restoration algorithms to locally remove hard-coded subtitles and text watermarks at lossless resolution and intelligently fill in the covered areas.
video-subtitle-remover has accumulated more than 10,700 Stars and 1,369 Forks on GitHub.
video-subtitle-remover is maintained by developer YaoFANGUK and is written in Python.
VSR is fully implemented locally, requires no third-party API applications, and video footage does not need to be uploaded to external servers.
The output files generated by VSR after subtitle removal maintain the original resolution, rather than avoiding the subtitle area through downsampling or cropping.
When processing 1080p video with an RTX 3070-class GPU with 8GB of VRAM, the AI restoration inference speed is typically far below real-time, and one minute of video may require several minutes or even longer.
If VSR is run using only a CPU, the speed gap compared to a GPU can be as large as tens of times.
For users without a dedicated GPU, VSR can be run on cloud GPU instances such as platforms like Colab and Autodl.
VSR also supports text watermark removal from static images.
All Facts (8)
video-subtitle-remover (VSR) is an open-source project that uses AI restoration algorithms to locally remove hard-coded subtitles and text watermarks at lossless resolution and intelligently fill in the covered areas.
50%Unverifiedvideo-subtitle-remover is maintained by developer YaoFANGUK and is written in Python.
50%UnverifiedVSR is fully implemented locally, requires no third-party API applications, and video footage does not need to be uploaded to external servers.
50%UnverifiedThe output files generated by VSR after subtitle removal maintain the original resolution, rather than avoiding the subtitle area through downsampling or cropping.
50%UnverifiedWhen processing 1080p video with an RTX 3070-class GPU with 8GB of VRAM, the AI restoration inference speed is typically far below real-time, and one minute of video may require several minutes or even longer.
50%UnverifiedIf VSR is run using only a CPU, the speed gap compared to a GPU can be as large as tens of times.
50%UnverifiedFor users without a dedicated GPU, VSR can be run on cloud GPU instances such as platforms like Colab and Autodl.
50%UnverifiedVSR also supports text watermark removal from static images.
50%