Optical Flow on MVSEC 1.0 (indoor_flying3)
0.4EPEScaleEvent
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ScaleEventBackbone=ViT-S/16, Protocol=Event Pretraining + Fully-Supervised2026.03 | 0.4 | 0.001 | |
| ECDDPBackbone=Swin-T/7, Protocol=Event Pretraining + Fully-Supervised2026.03 | 0.42 | 0.001 | |
| STPBackbone=Swin-T/7, Protocol=Event Pretraining + Fully-Supervised2026.03 | 0.43 | 0.001 | |
| EDCFlow (+in2)dt=1, Input=E, Learning paradigm=SL, training_inclusion=flying22025.06 | 0.49 | 0.0034 | |
| MultiCM (Burgers')dt=1, Input=E, Learning paradigm=MB2025.06 | 0.5 | 0.0028 | |
| EDCFlow (+in1)dt=1, Input=E, Learning paradigm=SL, training_inclusion=flying12025.06 | 0.54 | 0.0096 | |
| EV-MGRFlowNetdt=1, Input=E, Learning paradigm=USL2025.06 | 0.59 | 0.0129 | |
| ECDDPBackbone=ViT-S/16, Protocol=Event Pretraining + Fully-Supervised2026.03 | 0.61 | 0.08 | |
| Brebion et al.dt=1, Input=E, Learning paradigm=MB2025.06 | 0.71 | 0.021 | |
| STE-FlowNetdt=1, Input=E, Learning paradigm=SSL2025.06 | 0.72 | 0.013 | |
| DCEIFlowdt=1, Input=E+I1, Learning paradigm=SL2025.06 | 0.8 | 0.0177 | |
| Zhu et al.dt=1, Input=E, Learning paradigm=USL2025.06 | 0.87 | 0.03 | |
| STPBackbone=ViT-S/16, Protocol=Event Pretraining + Fully-Supervised2026.03 | 0.93 | 3.03 | |
| ECDPBackbone=ViT-S/16, Protocol=Event Pretraining + Fully-Supervised2026.03 | 1 | 3.11 | |
| ECDPBackbone=ResNet-50, Protocol=Event Pretraining + Fully-Supervised2026.03 | 1.12 | 5.26 | |
| DCEFlowProtocol=RGB initialization + Fully-Supervised2026.03 | 1.13 | 5.29 | |
| EDCFlowdt=1, Input=E, Learning paradigm=SL2025.06 | 1.38 | 0.0898 | |
| ESTMeasurement=time stamps, Kernel=exponential2019.04 | 1.4 | 9.44 | |
| ESTMeasurement=time stamps, Kernel=alpha2019.04 | 1.41 | 8.32 | |
| ESTMeasurement=time stamps, Kernel=learnt2019.04 | 1.43 | 6.47 | |
| Voxel GridMeasurement=polarity, Kernel=trilinear2019.04 | 1.45 | 11.4 | |
| EDCFlow (+in2)dt=4, Input=E, Learning paradigm=SL, training_inclusion=flying22025.06 | 1.47 | 0.0797 | |
| Voxel GridMeasurement=time stamps, Kernel=trilinear2019.04 | 1.5 | 12 | |
| IDNet-4dt=1, Input=E, Learning paradigm=SL, re-implemented=true2025.06 | 1.51 | 0.0958 | |
| ESTMeasurement=time stamps, Kernel=trilinear2019.04 | 1.51 | 8.29 | |
| EV-FlowNetdt=1, Input=E, Learning paradigm=SSL2025.06 | 1.53 | 0.119 | |
| EV-FlowNetMeasurement=count, Kernel=trilinear2019.04 | 1.53 | 11.9 | |
| TMAdt=1, Input=E, Learning paradigm=SL2025.06 | 1.58 | 0.2326 | |
| IDNet-8dt=1, Input=E, Learning paradigm=SL, re-implemented=true2025.06 | 1.59 | 0.1005 | |
| E-RAFTdt=1, Input=E, Learning paradigm=SL2025.06 | 1.66 | 0.252 | |
| EDCFlow (+in1)dt=4, Input=E, Learning paradigm=SL, training_inclusion=flying12025.06 | 1.67 | 0.1144 | |
| ESTBackbone=ResNet-18, Protocol=RGB initialization + Fully-Supervised2026.03 | 1.71 | 11.67 | |
| Event FrameMeasurement=time stamps, Kernel=trilinear2019.04 | 1.74 | 15.5 | |
| Two-Channel ImageMeasurement=time stamps, Kernel=trilinear2019.04 | 1.78 | 11.7 | |
| Two-Channel ImageMeasurement=count, Kernel=trilinear2019.04 | 1.84 | 17.7 | |
| MultiCM (Burgers')dt=4, Input=E, Learning paradigm=MB2025.06 | 2.06 | 0.1903 | |
| EV-MGRFlowNetdt=4, Input=E, Learning paradigm=USL2025.06 | 2.06 | 0.18 | |
| STE-FlowNetdt=4, Input=E, Learning paradigm=SSL2025.06 | 2.23 | 0.221 | |
| DCEIFlowdt=4, Input=E+I1, Learning paradigm=SL2025.06 | 2.51 | 0.2973 | |
| Zhu et al.dt=4, Input=E, Learning paradigm=USL2025.06 | 3.18 | 0.478 | |
| EV-FlowNetdt=4, Input=E, Learning paradigm=SSL2025.06 | 3.45 | 0.397 | |
| EDCFlowdt=4, Input=E, Learning paradigm=SL2025.06 | 3.48 | 0.3654 | |
| TMAdt=4, Input=E, Learning paradigm=SL2025.06 | 3.6 | 0.4202 | |
| IDNet-4dt=4, Input=E, Learning paradigm=SL, re-implemented=true2025.06 | 4.33 | 0.5291 | |
| E-RAFTdt=4, Input=E, Learning paradigm=SL2025.06 | 4.46 | 0.5711 | |
| IDNet-8dt=4, Input=E, Learning paradigm=SL, re-implemented=true2025.06 | 5.47 | 0.6858 |