Questions about the inference of the model/project? (DCVC-HEM)

[BeamReflection]

Thanks to all the authors for their hard work！

I download the pretrained models to compress/decompress my own video. I followed the README.md to transfer my video to images, and then run the test_video.py. I got a output.json file, but where is the output of the compressed video?

Also, I am confused about the inference part. It seems that the README.md talks a lot about the training process, can you explain some inference part? Thank you very much.