Models Matter, So Does Training: An Empirical Study of CNNs for Optical Flow Estimation

14 Sep 2018Deqing SunXiaodong YangMing-Yu LiuJan Kautz

We investigate two crucial and closely related aspects of CNNs for optical flow estimation: models and training. First, we design a compact but effective CNN model, called PWC-Net, according to simple and well-established principles: pyramidal processing, warping, and cost volume processing... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK BENCHMARK
Optical Flow Estimation KITTI 2012 PWC-Net + ft - axXiv Average End-Point Error 1.5 # 2
Optical Flow Estimation KITTI 2015 PWC-Net + ft - axXiv Fl-all 7.72 # 4
Optical Flow Estimation Sintel-clean PWC-Net Average End-Point Error 3.45 # 5
Optical Flow Estimation Sintel-final PWC-Net Average End-Point Error 4.6 # 8

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet