Optical Flow using deep learning


Are there any projects on pytorch that compute optical flow? This would probably be based on current research, where optical flow is computed using deep learning. I have not been able to find any pytorch code related to this, nor any pretrained models regarding this issue.

I think you mean the flownet model. It has quite some pytorch implementation


You can use the latest RAFT published on ECCV, the model currently performs best on the sintel dataset

Hi there, I’m interested too in optical flow, but I’m looking for a solution that is able to compute optical flow in real-time (i.e. I have an input video stream and I need to compute optical flow as the frames arrive), does anyone know anything about? Or maybe suggestions/workarounds.
All the solutions that I found compute optical flow as a pre-processing step, which is not suitable for my case.
Thank you :slight_smile: