Awesome-Optical-Flow

Why we included this project

Optical flow work tends to scatter across conference papers, so a single index that points each paper back to its official code saves real time. This list sorts the field by training paradigm, which means you can jump straight to supervised models like RAFT and GMFlow or to unsupervised, semi-supervised, and multi-frame approaches without wading through everything else. It also covers data synthesis, useful when you need to build or expand training sets. Every entry records the venue and year, so you can tell how recent a method is before digging into it. For anyone scoping a video understanding or tracking project, it reads as a map of the field rather than a random pile of links.