List your company
ResearchFormerly WizWand

Optical Flow on Middlebury (train)

0.22AEEClassic+NLP
0.16120.55810.9551.3519Nov 3, 2016May 15, 2017Nov 24, 2017Jun 5, 2018Dec 15, 2018Jun 26, 2019Jan 6, 2020
Updated 8mo ago

Evaluation results

MethodLinks
Classic+NLPCategory=Conventional2018.050.22
Classic+NLPFine-tuned=false, Time (s)=1022016.110.22
DeepFlow+backbone=CPU, runtime=51,940ms, improved_option=true2016.120.25
DeepFlowCategory=Conventional2018.050.25
FlowFieldsbackbone=CPU, runtime=22,810ms, improved_option=true2016.120.27
FlowFieldsCategory=Conventional2018.050.27
LiteFlowNet-ftCategory=Lightweight CNN, fine-tuned=true2018.050.3
EpicFlow+backbone=CPU, runtime=42,600ms, improved_option=true2016.120.31
EpicFlowCategory=Conventional2018.050.31
SPyNetCategory=Lightweight CNN2018.050.33
SPyNet-ftCategory=Lightweight CNN, fine-tuned=true2018.050.33
SPyNetFine-tuned=false, Time (s)=0.0692016.110.33
SPyNetFine-tuned=true, Time (s)=0.0692016.110.33
FlowNet2backbone=GPU, runtime=123ms, improved_option=true2016.120.35
FlowNet2-ft-sintelbackbone=GPU, runtime=123ms, improved_option=true, fine-tuned=Sintel2016.120.35
FlowNet2Category=Heavyweight CNN2018.050.35
FlowNet2-ft-sintelCategory=Heavyweight CNN, fine-tuned=Sintel2018.050.35
LDOF (GPU)backbone=GPU, runtime=6,270ms, improved_option=true2016.120.36
FlowNet2-CSS-ft-sdbackbone=GPU, runtime=69ms, improved_option=true2016.120.38
LiteFlowNetCategory=Lightweight CNN2018.050.39
FlowNet2-css-ft-sdbackbone=GPU, runtime=31ms, improved_option=true2016.120.43
LDOF (CPU)backbone=CPU, runtime=64,900ms, improved_option=true2016.120.44
FlowNet2-CSSbackbone=GPU, runtime=69ms, improved_option=true2016.120.44
LDOFCategory=Conventional2018.050.44
LiteFlowNetXCategory=Lightweight CNN2018.050.45
FlowNet2-cssbackbone=GPU, runtime=31ms, improved_option=true2016.120.54
FlowNet2-ft-kittibackbone=GPU, runtime=123ms, improved_option=true, fine-tuned=KITTI2016.120.56
FlowNet2-ft-kittiCategory=Heavyweight CNN, fine-tuned=KITTI2018.050.56
PCA-Layersbackbone=GPU, runtime=3,300ms, improved_option=true2016.120.66
PCA-LayersCategory=Conventional2018.050.66
FlowNet2-ssbackbone=GPU, runtime=14ms, improved_option=true2016.120.68
PCA-Flowbackbone=CPU, runtime=140ms, improved_option=true2016.120.7
SpyNetScale=4x2020.010.81
DIS-Fastbackbone=CPU, runtime=70ms, improved_option=true2016.120.92
FlowNetC-ftCategory=Heavyweight CNN, fine-tuned=true2018.050.93
FlowNetCFine-tuned=true, Time (s)=0.1502016.110.93
FlowNetS-ftCategory=Heavyweight CNN, fine-tuned=true2018.050.98
FlowNet2-CCategory=Heavyweight CNN2018.050.98
FlowNetSFine-tuned=true, Time (s)=0.0802016.110.98
FlowNet2-SCategory=Heavyweight CNN2018.051.04
FlowNetSbackbone=GPU, runtime=18ms, improved_option=true2016.121.09
FlowNetSCategory=Heavyweight CNN2018.051.09
FlowNetSFine-tuned=false, Time (s)=0.0802016.111.09
FlowNet2-S re-trainedCategory=Heavyweight CNN, re-trained=true2018.051.13
FlowNetCbackbone=GPU, runtime=32ms, improved_option=true2016.121.15
FlowNetCCategory=Heavyweight CNN2018.051.15
FlowNetCFine-tuned=false, Time (s)=0.1502016.111.15
FlowNet-SScale=4x2020.011.18
FlowNet2-sbackbone=GPU, runtime=7ms, improved_option=true2016.121.27
OFRnetFlow type=Super-resolved, Scale=4x2020.011.3
Upsampled FlowFlow type=Upsampled, Scale=4x2020.011.69