FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks

The FlowNet demonstrated that optical flow estimation can be cast as a learning problem. However, the state of the art with regard to the quality of the flow has still been defined by traditional methods. Particularly on small displacements and real-world data, FlowNet cannot compete with variational methods. In this paper, we advance the concept of end-to-end learning of optical flow and make it work really well. The large improvements in quality and speed are caused by three major contributions: first, we focus on the training data and show that the schedule of presenting data during training is very important. Second, we develop a stacked architecture that includes warping of the second image with intermediate optical flow. Third, we elaborate on small displacements by introducing a sub-network specializing on small motions. FlowNet 2.0 is only marginally slower than the original FlowNet but decreases the estimation error by more than 50%. It performs on par with state-of-the-art methods, while running at interactive frame rates. Moreover, we present faster variants that allow optical flow computation at up to 140fps with accuracy matching the original FlowNet.

PDF Abstract CVPR 2017 PDF CVPR 2017 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Dense Pixel Correspondence Estimation HPatches FlowNet2 Viewpoint I AEPE 5.99 # 5
Viewpoint II AEPE 15.55 # 5
Viewpoint III AEPE 17.09 # 5
Viewpoint IV AEPE 22.13 # 5
Viewpoint V AEPE 30.68 # 5
Skeleton Based Action Recognition JHMDB Pose Tracking FlowNet2 PCK@0.1 45.2 # 2
PCK@0.2 62.9 # 3
PCK@0.3 73.5 # 3
PCK@0.4 80.6 # 3
PCK@0.5 85.5 # 3
Optical Flow Estimation KITTI 2015 (train) FlowNet2 F1-all 30.0 # 13
EPE 10.08 # 11
Optical Flow Estimation Sintel-clean FlowNet2 Average End-Point Error 3.96 # 14


No methods listed for this paper. Add relevant methods here