RapidVideOCR ¶

Power by CPU [1]:

pip install rapid_videocr
rapid_videocr -o srt -i <rgb_images_dir> -s _output

Power by GPU [2][3]:

git clone --depth=1 https://github.com/SWHL/RapidVideOCR
uv venv
.venv\Scripts\activate.bat
uv pip install paddlepaddle-gpu==3.0.0b1 -i https://www.paddlepaddle.org.cn/packages/stable/cu123/
uv pip install get-pypi-latest-version
python setup.py install
# pip uninstall onnxruntime
# pip install onnxruntime-directml
uv pip install rapidocr_paddle
rapid_videocr --use_cuda -o srt -i <rgb_images_dir> -s _output

Cross-reference¶

videosubfinder.md