A pre-built training environment for Freyja subtitle OCR trainer.
This docker image contains:
-
compiled binaries of vapoursynth, ffms2 and fmtconv for training data reading.
-
CUDA and cuDNN for GPU accelerated training.
-
Jupyter Notebook and TensorBoard for easy remote access and training progress monitoring.
-
Other required python packages for training.
After pulling images from https://hub.docker.com/r/arition/freyja-sub-ocr, set up port forwarding on 8888:8888 for docker, and run following commands in container:
git clone https://github.com/freyjaSubOCR/freyja-sub-ocr-training # clone training repo
socat TCP-LISTEN:8888,bind=0.0.0.0,reuseaddr,fork TCP:127.0.0.1:8889 & # forward localhost jupyter instance to all interface
jupyter lab # start jupyter instance
Now you can access Jupyter instance on http://localhost:8888