List your company
ResearchFormerly WizWand

Optical Flow on Sintel Final (train)

0.66EPEVideoFlow
0.46841.76173.0554.3483Mar 26, 2020Apr 2, 2021Apr 9, 2022Apr 17, 2023Apr 23, 2024Apr 30, 2025May 8, 2026
Updated 5mo ago

Evaluation results

MethodLinks
VideoFlowSupervision=Supervised, Multi-frame data usage=true2026.040.66
FlowFormerSupervision=Supervised2026.040.74
SEA-RAFT (S)Extra Data=TartanAir, #refine=42026.050.782
GMFlowExtra Data=None, #refine=12026.051.11
OursExtra Data=None, #refine=02026.051.14
HD3Training Data=C + T + S/K2020.031.17
RAFT (2-view)Training Data=C + T + S/K2020.031.2
RAFTExtra Data=None, #refine=322026.051.217
RAFT (2-view)Training Data=C + T + S + K + H2020.031.22
RAFT (warm-start)Training Data=C + T + S + K + H2020.031.27
RAFTSupervision=Supervised2026.041.27
GMFlowExtra Data=None, #refine=02026.051.276
FlowSeek (T)Extra Data=TartanAir, #refine=42026.051.28
LiteFlowNet2Training Data=C + T + S + K + H2020.031.62
SelFlow-ftSupervision Type=Supervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Labeled evaluation data2020.061.77
DDVMZero-shot=true, Training Dataset=AF+FT+KU+TA2023.062
FlowNet2-ftSupervision Type=Supervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Labeled evaluation data2020.062.01
FlowNet2Training Data=C + T + S/K2020.032.01
PWC-Net-ftSupervision Type=Supervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Labeled evaluation data2020.062.21
PWC-Net+ftLearning paradigm=Supervised, Fine-tuning=true, Unsupervised training with test images=true2020.122.21
LiteFlowNet+ftLearning paradigm=Supervised, Fine-tuning=true, Unsupervised training with test images=true2020.122.23
VCN-ftSupervision Type=Supervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Labeled evaluation data2020.062.24
VCNTraining Data=C + T + S + K + H2020.032.24
RAFTZero-shot=true, Training Dataset=AF+FT+KU+TA2023.062.28
U2Flow (+FF)Supervision=Unsupervised, Multi-frame data usage=true, Bidirectional flow fusion (+FF)=true2026.042.29
DistractFlowLearning Setting=Supervised, Base Model=FlowFormer, Labeled data=C+T2023.032.31
DistractFlowLearning Setting=Semi-Supervised, Base Model=GMA, Labeled data=C+T, Unlabeled data=S/K2023.032.31
OCAILabeled data=C + T, Unlabeled data=S/K2024.032.32
U2FlowSupervision=Unsupervised2026.042.32
DistractFlowLearning Setting=Semi-Supervised, Base Model=FlowFormer, Labeled data=C+T, Unlabeled data=S/K2023.032.33
PWC-Net+Training Data=C + T + S + K + H2020.032.34
PWC-Net+Supervision=Supervised2026.042.34
DistractFlowLearning Setting=Semi-Supervised, Base Model=RAFT, Labeled data=C+T, Unlabeled data=S/K2023.032.35
DistractFlowLabeled data=C + T, Unlabeled data=S/K2024.032.35
RealFlowLabeled data=C + T, Unlabeled data=S/K2024.032.38
FlowFormerLearning Setting=Supervised, Labeled data=C+T2023.032.39
FlowFormerZero-shot=true, Training Dataset=Chairs -> Things2023.062.4
RAFTZero-shot=true, Training Dataset=AutoFlow2023.062.41
Perceiver IOZero-shot=true, Training Dataset=AutoFlow2023.062.42
FlowSupervisorLearning Setting=Semi-Supervised, Labeled data=C+T, Unlabeled data=S/K2023.032.46
FlowSupervisorLabeled data=C + T, Unlabeled data=T (subsampled)2024.032.46
IRR-PWCTraining Data=C + T + S/K2020.032.51
IRR-PWC+ftLearning paradigm=Supervised, Fine-tuning=true, Unsupervised training with test images=true2020.122.51
RAFT-ALearning Setting=Semi-Supervised, Labeled data=AutoFlow2023.032.57
RAFT-ALabeled data=AutoFlow [39]2024.032.57
Fixed TeacherLearning Setting=Semi-Supervised, Labeled data=C+T, Unlabeled data=S/K2023.032.58
Fixed TeacherLabeled data=C + T, Unlabeled data=T (subsampled)2024.032.58
SMURFSupervision=Unsupervised, Multi-frame data usage=true2026.042.58
DistractFlowLearning Setting=Supervised, Base Model=RAFT, Labeled data=C+T2023.032.61
DistractFlowLearning Setting=Supervised, Base Model=GMA, Labeled data=C+T2023.032.66
UPFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.122.67
RAFT-OCTCLearning Setting=Semi-Supervised, Labeled data=T (subsampled)2023.032.67
RAFT-OCTCLabeled data=C + T, Unlabeled data=T (subsampled)2024.032.67
UPFlowSupervision=Unsupervised2026.042.67
RAFT (2-view)Training Data=C + T2020.032.71
RAFT-SupervisedLabeled data=C + T2024.032.71
RAFTLearning Setting=Supervised, Labeled data=C+T2023.032.73
SMURFTrain on=Sintel2021.052.8
RAFTZero-shot=true, Training Dataset=Chairs -> Things2023.062.8
RAFTLearning paradigm=Supervised2020.122.83
GMALearning Setting=Supervised, Labeled data=C+T2023.032.85
UnSAMFlowSupervision=Unsupervised, Semantic inputs=true2026.043.07
M2FlowSupervision=Unsupervised, Multi-frame data usage=true2026.043.12
FlowNet2Supervision Type=Supervised2020.063.14
RAFT (small)Training Data=C + T2020.033.35
SMURFTrain on=Chairs2021.053.35
UFlow-trainSupervision Type=Unsupervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Unlabeled evaluation data2020.063.39
UFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.39
UFlowSupervision=Unsupervised2026.043.39
FlowNet2Training Data=C + T2020.033.54
SimFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.57
STFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.59
MaskFlowNetTraining Data=C + T2020.033.61
VCNSupervision Type=Supervised2020.063.62
VCNTraining Data=C + T2020.033.68
ARFlowSupervision=Unsupervised2026.043.73
LiteFlowNet2Training Data=C + T2020.033.78
NLFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.85
SelFlowSupervision Type=Unsupervised, Additional Modalities (MDM, SDM, MF)=Multi-frame flow (MF), Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Unlabeled evaluation data2020.063.87
SelFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.87
ARFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.87
SelFlowSupervision=Unsupervised, Multi-frame data usage=true2026.043.87
PWC-NetSupervision Type=Supervised2020.063.93
PWC-NetTraining Data=C + T2020.033.93
PWC-NetLearning paradigm=Supervised2020.123.93
DDFlowSupervision Type=Unsupervised, Training Domain Condition (Labeled/Unlabeled/Related evaluation data)=Unlabeled evaluation data2020.063.98
DDFlowTrain on=Sintel2021.053.98
DDFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.123.98
LiteFlowNetTraining Data=C + T2020.034.04
LiteFlowNetLearning paradigm=Supervised2020.124.05
UFlow-testSupervision Type=Unsupervised2020.064.09
UFlowTrain on=Sintel2021.054.09
SpyNet+ftLearning paradigm=Supervised, Fine-tuning=true, Unsupervised training with test images=true2020.124.32
FlowNetS+ftLearning paradigm=Supervised, Fine-tuning=true, Unsupervised training with test images=true2020.124.44
SMURFTrain on=KITTI2021.054.47
DDFlowTrain on=Chairs2021.054.85
EpiFlowLearning paradigm=Unsupervised, Unsupervised training with test images=true2020.124.99
EPIFlowSupervision Type=Unsupervised2020.065.08
UFlowTrain on=Chairs2021.055.12
FlowNetSLearning paradigm=Supervised2020.125.45