Video Retrieval on HMDB51 (split 1)
45.9Top-1 AccuracySCE
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SCEBackbone=ResNet3D-50, Resp=16, Tp=224^2, Rese=16, Te=256^2, Pretrain=K4002022.12 | 45.9 | 69.9 | 80.5 | — | — | |
| SCEBackbone=ResNet3D-50, Resp=8, Tp=224^2, Rese=8, Te=256^2, Pretrain=K4002022.12 | 43 | 67 | 79 | — | — | |
| ProViCoBackbone=ResNet3D-18, Resp=16, Tp=112^2, Pretrain=K4002022.12 | 40.1 | 60.6 | 75.2 | — | — | |
| SCEBackbone=ResNet3D-18, Resp=8, Tp=224^2, Rese=8, Te=256^2, Pretrain=K4002022.12 | 40.1 | 63.3 | 75.4 | — | — | |
| SCEBackbone=ResNet3D-18, Resp=16, Tp=112^2, Rese=16, Te=128^2, Pretrain=K4002022.12 | 37.8 | 62.1 | 73.8 | — | — | |
| ProViCoBackbone=ResNet3D-18, Resp=16, Tp=112^2, Pretrain=UCF1012022.12 | 35.9 | 55.2 | 74.3 | — | — | |
| Supervised SLICInput Size=32 x 128^2, Architecture=R3D-182022.06 | 33 | 57.4 | 70 | 82.7 | — | |
| CATEBackbone=ResNet3D-50, Resp=8, Tp=224^2, Rese=32, Te=256^2, Pretrain=K4002022.12 | 33 | 56.8 | 69.4 | — | — | |
| TECBackbone=ResNet3D-18, Resp=16, Tp=128^2, Rese=16, Te=128^2, Pretrain=UCF1012022.12 | 32.2 | 60.3 | 71.6 | — | — | |
| ViCC-R+FBackbone=S3D, Modality=V2021.06 | 29.7 | 54.6 | 66 | 76.2 | — | |
| SLICInput Size=32 x 128^2, Architecture=R3D-182022.06 | 28.9 | 52.8 | 65.4 | 77.8 | — | |
| ViCC-R+FBackbone=R(2+1)D, Modality=V2021.06 | 28.3 | 52.7 | 65.3 | 77 | — | |
| SLICInput Size=32 x 128^2, Architecture=S3D-232022.06 | 26.8 | 52.9 | 66.2 | 78.1 | — | |
| Supervised SLICInput Size=16 x 128^2, Architecture=R3D-182022.06 | 26.8 | 51.2 | 65.1 | 76.3 | — | |
| CoCLRInput Size=32 x 128^2, Architecture=S3D-23, Input Modality=RGB + Optical Flow2022.06 | 26.1 | 45.8 | 57.9 | 69.7 | — | |
| CoCLRYear=2020, Backbone=S3D, Modality=V2021.06 | 26.1 | 45.8 | 57.9 | 69.7 | — | |
| TransRankBackbone=ResNet3D-18, Resp=16, Tp=112^2, Pretrain=K2002022.12 | 25.5 | 52.3 | 65.8 | — | — | |
| ViCC-RGBBackbone=S3D, Modality=V2021.06 | 25.5 | 49.6 | 61.9 | 72.5 | — | |
| SLICInput Size=16 x 128^2, Architecture=R3D-182022.06 | 25.3 | 49.8 | 64.9 | 76.1 | — | |
| ViCC-RGBBackbone=R(2+1)D, Modality=V2021.06 | 25.3 | 50.4 | 64 | 77.5 | — | |
| ViCC-R+FBackbone=R3D, Modality=V2021.06 | 25.2 | 48.1 | 61.1 | 72.7 | — | |
| CoCLRInput Size=32 x 128^2, Architecture=S3D-23, Input Modality=RGB2022.06 | 23.2 | 43.2 | 53.5 | 65.5 | — | |
| TCLRInput Size=16 x 112^2, Architecture=R3D-18, Input Modality=RGB2022.06 | 22.8 | 45.4 | 57.8 | 73.1 | — | |
| TCLRBackbone=ResNet3D-18, Resp=16, Tp=112^2, Pretrain=UCF1012022.12 | 22.8 | 45.4 | 57.8 | — | — | |
| ViCCBackbone=ResNet3D-18, Resp=16, Tp=128^2, Rese=16, Te=128^2, Pretrain=UCF1012022.12 | 22.7 | 46.2 | 60.9 | — | — | |
| ViCC-RGBBackbone=R3D, Modality=V2021.06 | 22.7 | 46.2 | 60.9 | 74.1 | — | |
| SLICInput Size=16 x 128^2, Architecture=S3D-232022.06 | 21.8 | 46.3 | 59.7 | 72.4 | — | |
| MFOBackbone=ResNet3D-18, Resp=16, Tp=112^2, Rese=16, Te=112^2, Pretrain=K4002022.12 | 20.7 | 40.8 | 55.2 | — | — | |
| IICInput Size=16 x 112^2, Architecture=R3D-18, Input Modality=RGB + Optical Flow2022.06 | 19.7 | 42.9 | 57.1 | 70.6 | — | |
| TransRankBackbone=ResNet3D-18, Resp=16, Tp=112^2, Pretrain=UCF1012022.12 | 19.4 | 45.4 | 59.1 | — | — | |
| Supervised SLICInput Size=32 x 128^2, Architecture=S3D-232022.06 | 19.1 | 45.1 | 58 | 70.7 | — | |
| Supervised SLICInput Size=16 x 128^2, Architecture=S3D-232022.06 | 17.3 | 39.4 | 53.7 | 67.7 | — | |
| IICInput Size=16 x 112^2, Architecture=R3D-18, Input Modality=RGB2022.06 | 13.4 | 32.7 | 46.7 | 61.5 | — | |
| VCPBackbone=C3D2020.01 | 7.8 | 23.8 | 35.3 | 49.3 | 71.6 | |
| MemDPCInput Size=40 x 128^2, Architecture=R3D-18, Input Modality=RGB2022.06 | 7.7 | 25.7 | 40.6 | 57.7 | — | |
| MemDPCBackbone=ResNet3D-18, Resp=40, Tp=224^2, Rese=40, Te=224^2, Pretrain=UCF1012022.12 | 7.7 | 25.7 | 40.6 | — | — | |
| MemDPCYear=2020, Backbone=R2D3D, Modality=V2021.06 | 7.7 | 25.7 | 40.6 | 57.7 | — | |
| VCOPBackbone=R3D2020.01 | 7.6 | 22.9 | 34.4 | 48.8 | 68.9 | |
| VCPBackbone=R3D2020.01 | 7.6 | 24.4 | 36.3 | 53.6 | 76.4 | |
| VCPInput Size=16 x 112^2, Architecture=R3D-18, Input Modality=RGB2022.06 | 7.6 | 24.4 | 36.3 | 53.6 | — | |
| VCOPInput Size=16 x 112^2, Architecture=R3D-18, Input Modality=RGB2022.06 | 7.6 | 22.9 | 34.4 | 48.8 | — | |
| VCOPYear=2019, Backbone=R3D, Modality=V2021.06 | 7.6 | 22.9 | 34.4 | 48.8 | — | |
| random initializationBackbone=C3D2020.01 | 7.4 | 20.5 | 31.9 | 44.5 | 66.3 | |
| VCOPBackbone=C3D2020.01 | 7.4 | 22.6 | 34.4 | 48.5 | 70.1 | |
| random initializationBackbone=R3D2020.01 | 6.7 | 18.3 | 28.3 | 43.1 | 67.9 | |
| VCPBackbone=R(2+1)D2020.01 | 6.7 | 21.3 | 32.7 | 49.2 | 73.3 | |
| VCPYear=2020, Backbone=R(2+1)D, Modality=V2021.06 | 6.7 | 21.3 | 32.7 | 49.2 | — | |
| VCOPBackbone=R(2+1)D2020.01 | 5.7 | 19.5 | 30.7 | 45.8 | 67 | |
| random initializationBackbone=R(2+1)D2020.01 | 4.5 | 14.8 | 23.4 | 38.9 | 63 |