List your company
ResearchFormerly WizWand

Optical Flow on MPI Sintel (train)

0.74EPE (Final)FlowFormer
0.45322.38914.3256.2609Nov 11, 2022Feb 9, 2023May 10, 2023Aug 8, 2023Nov 6, 2023Feb 4, 2024May 4, 2024
Updated 8mo ago

Evaluation results

MethodLinks
FlowFormerLearning Paradigm=Supervised, Parameters=18.2M, Evaluation data used in training=true2024.050.740.48
RAFT-ftLearning Setting=Supervised, Training Data=S/K, Fine-tuned=true2022.111.20.77
RAFTLearning Paradigm=Supervised, Parameters=5.3M, Evaluation data used in training=true2024.051.270.77
FlowNet2-ftLearning Setting=Supervised, Training Data=S/K, Fine-tuned=true2022.112.011.45
PWC-Net-ftLearning Setting=Supervised, Training Data=S/K, Fine-tuned=true2022.112.211.7
PWC-Net+Learning Paradigm=Supervised, Parameters=8.8M, Evaluation data used in training=true2024.052.341.71
CamLiRAFTTraining Data=T, Input=RGB + XYZ2023.032.381.27
IRR-PWCLearning Paradigm=Supervised, Parameters=6.4M, Evaluation data used in training=true2024.052.511.92
CamLiPWCTraining Data=T, Input=RGB + XYZ2023.032.551.79
S-Flow2024.012.591.3
Ef-RAFTFine-tuned=false2024.012.61.27
UPFlowLearning Setting=Unsupervised, Training Data=S/Kraw2022.112.672.33
EMD-S2024.012.671.31
UPFlowLearning Paradigm=Unsupervised, Parameters=3.5M, Evaluation data used in training=true2024.052.672.33
KPA-Flow2024.012.681.28
AGFlow2024.012.691.31
FastFlowNet-ftLearning Setting=Supervised, Training Data=S/K, Fine-tuned=true2022.112.712.08
RAFTLearning Setting=Supervised, Training Data=C+T2022.112.711.43
RAFT2024.012.711.43
GMFlowNet2024.012.711.14
RAFTTraining Data=C + T, Input=RGB2023.032.711.43
GMA2024.012.741.3
CRAFT2024.012.791.27
ASFlowLearning Setting=Unsupervised, Training Data=S/Kraw2022.112.892.4
RAFT-3DTraining Data=T, Input=RGB + Depth2023.032.911.75
FM-RAFT2024.012.951.29
UnSAMFlow (+aug +hg +mf)Learning Paradigm=Unsupervised, Parameters=2.6M, Evaluation data used in training=true, semantic augmentation module=true, homography smoothness loss=true, mask feature module=true, SAM used in training=true, SAM used in inference=true2024.053.072.21
UnSAMFlow (+aug +hg)Learning Paradigm=Unsupervised, Parameters=2.5M, Evaluation data used in training=true, semantic augmentation module=true, homography smoothness loss=true, SAM used in training=true2024.053.12.25
FlowNet2Learning Setting=Supervised, Training Data=C+T2022.113.142.02
MDFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.142.17
Flow1D2024.013.271.98
UnSAMFlow (+aug)Learning Paradigm=Unsupervised, Parameters=2.5M, Evaluation data used in training=true, semantic augmentation module=true, SAM used in training=true2024.053.332.35
OIFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.352.44
UFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.392.5
UFlowLearning Paradigm=Unsupervised, Evaluation data used in training=true2024.053.392.5
MDFlow-FastLearning Setting=Unsupervised, Training Data=C + S/K2022.113.472.53
FlowNet22024.013.542.02
FlowNet2Training Data=C + T, Input=RGB2023.033.542.02
SimFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.572.86
SimFlowLearning Paradigm=Unsupervised, Evaluation data used in training=true2024.053.572.86
STFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.592.91
MaskFlowNet2024.013.612.25
UnSAMFlow (baseline)Learning Paradigm=Unsupervised, Parameters=2.5M, Evaluation data used in training=true2024.053.632.67
VCN2024.013.682.21
DistillFlowLearning Setting=Unsupervised, Training Data=S + Sraw/Kraw2022.113.72.61
ARFlowLearning Setting=Unsupervised, Training Data=Sraw/Kraw2022.113.732.79
ARFlowLearning Paradigm=Unsupervised, Parameters=2.2M, Evaluation data used in training=true2024.053.732.79
DICL2024.013.771.94
LiteFlowNet22024.013.782.24
SelFlowLearning Setting=Unsupervised, Training Data=Sraw/K2022.113.872.88
SelFlowLearning Paradigm=Unsupervised, Parameters=4.8M, Evaluation data used in training=true2024.053.872.88
PWC-NetLearning Setting=Supervised, Training Data=C+T2022.113.932.55
PWC-Net2024.013.932.55
PWC-NetTraining Data=C + T, Input=RGB2023.033.932.55
DDFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.113.982.92
DDFlowLearning Paradigm=Unsupervised, Parameters=4.3M, Evaluation data used in training=true2024.053.982.92
LiteFlowNet2024.014.042.48
FastFlowNetLearning Setting=Supervised, Training Data=C+T2022.114.142.89
EPIFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.114.993.54
MFOccFlowLearning Setting=Unsupervised, Training Data=R + S/K2022.115.523.89
OccAwareFlowLearning Setting=Unsupervised, Training Data=C + S/K2022.115.954.03
UnFlow-CSSLearning Setting=Unsupervised, Training Data=SY/Kraw2022.117.91—
UnFlow-CSSLearning Paradigm=Unsupervised, Parameters=116.6M2024.057.91—