MaskFlownet: Asymmetric Feature Matching with Learnable Occlusion Mask

Benchmark Model Rank Results
optical-flow-estimation-on-kitti-2012MaskFlownet-S#2Average End-Point Error: 1.1
optical-flow-estimation-on-kitti-2012MaskFlownet#3Average End-Point Error: 1.1
optical-flow-estimation-on-kitti-2015MaskFlownet#7Fl-all: 6.11
optical-flow-estimation-on-kitti-2015MaskFlownet-S#9Fl-all: 6.81
optical-flow-estimation-on-kitti-2015-trainMaskFlowNet#14F1-all: 23.1
optical-flow-estimation-on-sintel-cleanMaskFlownet#14Average End-Point Error: 2.52
optical-flow-estimation-on-sintel-cleanMaskFlownet-S#15Average End-Point Error: 2.77
optical-flow-estimation-on-sintel-finalMaskFlownet#14Average End-Point Error: 4.17
optical-flow-estimation-on-sintel-finalMaskFlownet-S#16Average End-Point Error: 4.38