Video Retrieval on UCF101 (test)
98Top-1 AccHuman estimate
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Human estimateDescription Variant=Object2024.06 | 98 | — | — | — | — | — | — | — | — | |
| Human estimateDescription Variant=Masked Object2024.06 | 98 | — | — | — | — | — | — | — | — | |
| Our MethodDescription Variant=Object2024.06 | 58.46 | — | — | — | — | — | — | — | — | |
| CoCLR-2StreamTraining Dataset=UCF101, Modality=Two-stream, Averaged similarity=true2020.10 | 55.9 | 70.8 | 76.9 | 82.5 | — | — | — | — | — | |
| MCNBackbone=R3D2021.08 | 53.8 | 70.2 | 78.3 | 83.4 | 89.7 | — | — | — | — | |
| CoCLR-RGBTraining Dataset=UCF101, Modality=RGB2020.10 | 53.3 | 69.4 | 76.6 | 82 | — | — | — | — | — | |
| CoCLR2021.08 | 53.3 | 69.4 | 76.6 | 82 | — | — | — | — | — | |
| ViFiCLIPDescription Variant=Object2024.06 | 52.7 | — | — | — | — | — | — | — | — | |
| MCNBackbone=R(2+1)D2021.08 | 52.5 | 69.5 | 77.9 | 83.1 | 89.3 | — | — | — | — | |
| XCLIPDescription Variant=Object2024.06 | 52.37 | — | — | — | — | — | — | — | — | |
| CoCLR-FlowTraining Dataset=UCF101, Modality=Flow2020.10 | 51.9 | 68.5 | 75 | 80.8 | — | — | — | — | — | |
| Text4VisDescription Variant=Object2024.06 | 51.23 | — | — | — | — | — | — | — | — | |
| Our MethodDescription Variant=Masked Object2024.06 | 47.8 | — | — | — | — | — | — | — | — | |
| InfoNCE-FlowTraining Dataset=UCF101, Modality=Flow2020.10 | 45.5 | 67.5 | 75.4 | 82.7 | — | — | — | — | — | |
| IICBackbone=R3D, Sampling strategy=shuffle, Views=RGB + Optical Flow (v)2020.08 | 42.4 | 60.9 | 69.2 | 77.1 | 86.5 | — | — | — | — | |
| IIC2021.08 | 42.4 | 60.9 | 69.2 | 77.1 | 86.5 | — | — | — | — | |
| IICBackbone=R3D, Sampling strategy=repeat, Views=RGB + Optical Flow (u)2020.08 | 41.8 | 60.4 | 69.5 | 78.4 | 87.7 | — | — | — | — | |
| IICBackbone=R3D, Sampling strategy=repeat, Views=RGB + Residual2020.08 | 36.5 | 54.1 | 62.9 | 72.4 | 83.4 | — | — | — | — | |
| InfoNCE-RGBTraining Dataset=UCF101, Modality=RGB2020.10 | 36 | 52 | 61.8 | 71 | — | — | — | — | — | |
| ViFiCLIPDescription Variant=Masked Object2024.06 | 34.76 | — | — | — | — | — | — | — | — | |
| IICBackbone=R3D, Sampling strategy=shuffle, Views=RGB + Residual2020.08 | 34.6 | 53 | 62.3 | 71.7 | 82.4 | — | — | — | — | |
| Text4VisDescription Variant=Masked Object2024.06 | 33.8 | — | — | — | — | — | — | — | — | |
| XCLIPDescription Variant=Masked Object2024.06 | 32.01 | — | — | — | — | — | — | — | — | |
| PP2021.08 | 31.9 | 49.7 | 59.2 | 68.9 | 80.2 | — | — | — | — | |
| Catch-the-PatchPre-training Dataset=Kinetics-400, Backbone=R3D-182021.05 | 29 | 47.3 | — | — | — | — | — | — | — | |
| CMCBackbone=R3D, Input=RGB + Optical Flow2020.08 | 26.4 | 37.7 | 45.1 | 53.2 | 66.3 | — | — | — | — | |
| CMC2021.08 | 26.4 | 37.7 | 45.1 | 53.2 | 66.3 | — | — | — | — | |
| One-viewBackbone=R3D, Input=RGB only2020.08 | 26.2 | 39.3 | 46.8 | 55.6 | 66.8 | — | — | — | — | |
| TempTransPre-training Dataset=Kinetics-400, Backbone=R3D-182021.05 | 26.1 | 48.5 | — | — | — | — | — | — | — | |
| Vanilla CLIPDescription Variant=Object2024.06 | 25.92 | — | — | — | — | — | — | — | — | |
| BüchlerBackbone=R3D2020.08 | 25.7 | 36.2 | 42.2 | 49.2 | 59.5 | — | — | — | — | |
| BuchlerDate=2018, Training Dataset=UCF1012020.10 | 25.7 | 36.2 | 42.2 | 49.2 | — | — | — | — | — | |
| Büchler2021.08 | 25.7 | 36.2 | 42.2 | 49.2 | 59.5 | — | — | — | — | |
| Catch-the-PatchPre-training Dataset=UCF-101, Backbone=R3D-182021.05 | 23.4 | 40.9 | — | — | — | — | — | — | — | |
| Vanilla CLIPDescription Variant=Masked Object2024.06 | 23.33 | — | — | — | — | — | — | — | — | |
| PRPPre-training Dataset=UCF-101, Backbone=R3D-182021.05 | 22.8 | 38.5 | — | — | — | — | — | — | — | |
| MemDPCDate=2020, Training Dataset=UCF1012020.10 | 20.2 | 40.4 | 52.4 | 64.7 | — | — | — | — | — | |
| OPNBackbone=R3D2020.08 | 19.9 | 28.7 | 34 | 40.6 | 51.6 | — | — | — | — | |
| OPNDate=2017, Training Dataset=UCF1012020.10 | 19.9 | 28.7 | 34 | 40.6 | — | — | — | — | — | |
| PacePre-training Dataset=UCF-101, Backbone=R3D-182021.05 | 19.9 | 36.2 | — | — | — | — | — | — | — | |
| OPN2021.08 | 19.9 | 28.7 | 34 | 40.6 | 51.6 | — | — | — | — | |
| VCP2021.08 | 19.9 | 33.7 | 42 | 50.5 | 64.4 | — | — | — | — | |
| JigsawBackbone=R3D2020.08 | 19.7 | 28.5 | 33.5 | 40 | 49.4 | — | — | — | — | |
| JigsawDate=2016, Training Dataset=UCF1012020.10 | 19.7 | 28.5 | 33.5 | 40 | — | — | — | — | — | |
| Jigsaw2021.08 | 19.7 | 28.5 | 33.5 | 40 | 49.4 | — | — | — | — | |
| VCPBackbone=R3D2020.08 | 18.6 | 33.6 | 42.5 | 53.5 | 68.1 | — | — | — | — | |
| VCPDate=2020, Training Dataset=UCF1012020.10 | 18.6 | 33.6 | 42.5 | 53.5 | — | — | — | — | — | |
| VCPPre-training Dataset=UCF-101, Backbone=R3D-182021.05 | 18.6 | 33.6 | — | — | — | — | — | — | — | |
| VCOPBackbone=R3D2020.08 | 14.1 | 30.3 | 40.4 | 51.1 | 66.5 | — | — | — | — | |
| VCOPDate=2019, Training Dataset=UCF1012020.10 | 14.1 | 30.3 | 40.4 | 51.1 | — | — | — | — | — | |
| VCOPPre-training Dataset=UCF-101, Backbone=R3D-182021.05 | 14.1 | 30.3 | — | — | — | — | — | — | — | |
| SpeedNetDate=2020, Training Dataset=K4002020.10 | 13 | 28.1 | 37.5 | 49.5 | — | — | — | — | — | |
| SpeedNetPre-training Dataset=Kinetics-400, Backbone=R3D-182021.05 | 13 | 28.1 | — | — | — | — | — | — | — | |
| VCOP2021.08 | 10.7 | 25.9 | 35.4 | 47.3 | 63.9 | — | — | — | — | |
| R3D (random)Backbone=R3D2020.08 | 9.9 | 18.9 | 26 | 35.5 | 51.9 | — | — | — | — | |
| CATEBackbone=R3D-50 (31.7M), Pretrain=K4002022.04 | — | — | — | — | — | 54.9 | 68.3 | 75.1 | 82.3 | |
| CentralizedPre-training dataset=K400 (Non-IID)2022.07 | — | — | — | — | — | 29 | 47.3 | — | — | |
| DSMBackbone=I3D (25.0M), Pretrain=UCF1012022.04 | — | — | — | — | — | 17.4 | 35.2 | 45.3 | 57.8 | |
| FedAvg (Baseline)Pre-training dataset=K400 (Non-IID)2022.07 | — | — | — | — | — | 32.62 | 50.41 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=0, Beta (β)=02022.07 | — | — | — | — | — | 34.34 | 51.71 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=1, Beta (β)=02022.07 | — | — | — | — | — | 34.23 | 52.21 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=0, Beta (β)=12022.07 | — | — | — | — | — | 35.61 | 52.18 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=1, Beta (β)=12022.07 | — | — | — | — | — | 35.66 | 52.34 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=0.9, Beta (β)=02022.07 | — | — | — | — | — | 35.5 | 54.27 | — | — | |
| FedVSSLPre-training dataset=K400 (Non-IID), Alpha (α)=0.9, Beta (β)=12022.07 | — | — | — | — | — | 35.34 | 52.34 | — | — | |
| MCNBackbone=R3D-18 (20.2M), Pretrain=UCF101, additional residual views=true2022.04 | — | — | — | — | — | 53.8 | 70.2 | 78.3 | 83.4 | |
| MFOBackbone=R3D-18 (20.2M), Pretrain=K4002022.04 | — | — | — | — | — | 41.5 | 60.6 | 71.2 | 80.1 | |
| MFOBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 39.6 | 57.6 | 69.2 | 78 | |
| PaceBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 23.8 | 38.1 | 46.4 | 56.6 | |
| ProViCoBackbone=R3D-18 (20.2M), Pretrain=K4002022.04 | — | — | — | — | — | 67.6 | 81.4 | 90.1 | 94.7 | |
| ProViCoBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 63.8 | 75.1 | 84.8 | 89.2 | |
| PRPBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 22.8 | 38.5 | 46.7 | 55.2 | |
| SpeedNetBackbone=S3D-G (9.1M), Pretrain=K4002022.04 | — | — | — | — | — | 13 | 28.1 | 37.5 | 49.5 | |
| STSBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 38.3 | 59.9 | 68.9 | 77.2 | |
| TCLRBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 56.2 | 72.2 | 79 | 85.3 | |
| TECBackbone=R3D-18 (20.2M), Pretrain=K4002022.04 | — | — | — | — | — | 66.9 | 83.1 | 88.8 | 93.3 | |
| TECBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 62.5 | 78.4 | 84.1 | 88.8 | |
| VCPBackbone=R3D-18 (20.2M), Pretrain=UCF1012022.04 | — | — | — | — | — | 18.6 | 33.6 | 42.5 | 53.5 |