DDFlow: Learning Optical Flow with Unlabeled Data Distillation
This addresses the problem of learning optical flow without labeled data for computer vision applications, representing a strong incremental improvement over prior unsupervised methods.
The paper tackles optical flow estimation from unlabeled data by using a data distillation approach that distills reliable predictions from a teacher network to guide a student network, achieving significantly higher accuracy than existing unsupervised methods on benchmarks like Flying Chairs, MPI Sintel, KITTI 2012, and KITTI 2015, with real-time performance.
We present DDFlow, a data distillation approach to learning optical flow estimation from unlabeled data. The approach distills reliable predictions from a teacher network, and uses these predictions as annotations to guide a student network to learn optical flow. Unlike existing work relying on hand-crafted energy terms to handle occlusion, our approach is data-driven, and learns optical flow for occluded pixels. This enables us to train our model with a much simpler loss function, and achieve a much higher accuracy. We conduct a rigorous evaluation on the challenging Flying Chairs, MPI Sintel, KITTI 2012 and 2015 benchmarks, and show that our approach significantly outperforms all existing unsupervised learning methods, while running at real time.