AI RESEARCH
Rethinking Dense Optical Flow without Test-Time Scaling
arXiv CS.CV
•
ArXi:2605.08000v1 Announce Type: new Recent progress in dense optical flow has been driven by increasingly complex architectures and multi-step refinement for test-time scaling. While these approaches achieve strong benchmark performance, they also require substantial computation during inference. This raises a fundamental question: Is scaling test-time computation the only way to improve dense optical flow accuracy? We argue that it is not.