Action Recognition on Charades (val)
63.6mAPLaIAR Oracle
Evaluation Results
| Method | Links | |
|---|---|---|
| LaIAR OracleBackbone=R101, Modality=RGB-Bbox2024.04 | 63.6 | |
| OR2G OracleBackbone=R101, Modality=RGB-SG2024.04 | 63.3 | |
| VideoMLN OracleBackbone=R101, Modality=RGB-SG2024.04 | 62.8 | |
| SGFB OracleBackbone=R101-I3D-NL, Modality=RGB-SG2024.04 | 60.3 | |
| SlowFast NLBackbone=R101-I3D-NL, Feature bank=false, Pretrain=K6002021.04 | 45.2 | |
| LaIARBackbone=R101-I3D-NL, Modality=RGB2024.04 | 45.1 | |
| OR2GBackbone=R101-I3D-NL, Modality=RGB2024.04 | 44.9 | |
| SGFBBackbone=R101-I3D-NL, Modality=RGB2024.04 | 44.3 | |
| JMRN + R101-NL-LFBFeatures=P + R2020.10 | 43.23 | |
| TQNBackbone=R101-I3D-NL, Feature bank=true, Pretrain=K4002021.04 | 42.9 | |
| PoTion + R101-NL-LFBFeatures=P + R2020.10 | 42.84 | |
| LFB NLbackbone=R101-I3D-NL, modality=RGB2018.12 | 42.5 | |
| SlowFast NLBackbone=R101-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 42.5 | |
| LFB NLBackbone=R101-I3D-NL, Feature bank=true, Pretrain=K4002021.04 | 42.5 | |
| R101-I3D-NL LFBFeatures=R2020.10 | 42.5 | |
| LFBBackbone=R101-I3D-NL, Modality=RGB2024.04 | 42.5 | |
| STO ens.backbone=R101-I3D-NL, modality=RGB2018.12 | 42.3 | |
| TQNBackbone=R101-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 42.1 | |
| SlowFastBackbone=R101, Modality=RGB2024.04 | 42.1 | |
| 3D CNN ens.backbone=R101-I3D-NL, modality=RGB2018.12 | 41.7 | |
| TQNBackbone=R50-I3D-NL, Feature bank=true, Pretrain=K4002021.04 | 41.4 | |
| STObackbone=R101-I3D-NL, modality=RGB2018.12 | 41 | |
| STOBackbone=R101-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 41 | |
| PA3D + GCN + I3D + NL-I3DFeatures=P + CF + PAF + R2020.10 | 41 | |
| LFB Maxbackbone=R101-I3D-NL, modality=RGB2018.12 | 40.9 | |
| LFB Avgbackbone=R101-I3D-NL, modality=RGB2018.12 | 40.8 | |
| TQNBackbone=R50-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 40.7 | |
| LFB NLbackbone=R50-I3D-NL, modality=RGB2018.12 | 40.3 | |
| 3D CNNbackbone=R101-I3D-NL, modality=RGB2018.12 | 40.3 | |
| LFB NLBackbone=R50-I3D-NL, Feature bank=true, Pretrain=K4002021.04 | 40.3 | |
| STO ens.backbone=R50-I3D-NL, modality=RGB2018.12 | 40 | |
| STRG [52]backbone=R101-I3D-NL, modality=RGB2018.12 | 39.7 | |
| STRGBackbone=R101-I3D-NL, Modality=RGB2024.04 | 39.7 | |
| STObackbone=R50-I3D-NL, modality=RGB2018.12 | 39.6 | |
| STOBackbone=R50-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 39.6 | |
| 3D CNN ens.backbone=R50-I3D-NL, modality=RGB2018.12 | 39.5 | |
| Slow-8×8FLOPs (G)=54.9, views=302021.05 | 39 | |
| LFB Maxbackbone=R50-I3D-NL, modality=RGB2018.12 | 38.6 | |
| LFB Avgbackbone=R50-I3D-NL, modality=RGB2018.12 | 38.4 | |
| VideoMLNBackbone=R101, Modality=RGB2024.04 | 38.4 | |
| 3D CNNbackbone=R50-I3D-NL, modality=RGB2018.12 | 38.3 | |
| R50-I3D-NLFeatures=R2020.10 | 38.29 | |
| I3D-NL [51]backbone=R101-I3D-NL, modality=RGB2018.12 | 37.5 | |
| STRG [52]backbone=R50-I3D-NL, modality=RGB2018.12 | 37.5 | |
| I3D-NLBackbone=R50-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 37.5 | |
| STRGBackbone=R50-I3D-NL, Feature bank=false, Pretrain=K4002021.04 | 37.5 | |
| NL I3DFeatures=R2020.10 | 37.5 | |
| GCNFeatures=R2020.10 | 36.2 | |
| I3D [5] (from [51])backbone=R101-I3D, modality=RGB2018.12 | 35.5 | |
| I3DBackbone=R101-I3D, Feature bank=false, Pretrain=K4002021.04 | 35.5 | |
| I3D-NL [51] (from [52])backbone=R50-I3D-NL, modality=RGB2018.12 | 33.5 | |
| I3D [5]backbone=Inception-I3D, modality=RGB2018.12 | 32.9 | |
| I3DFeatures=R2020.10 | 32.72 | |
| MultiScale TRN [58]backbone=Inception, modality=RGB2018.12 | 25.2 | |
| TRNmsBackbone=Inception, Feature bank=false, Pretrain=ImageNet2021.04 | 25.2 | |
| CoSlow_64FLOPs (G)=6.9, views=12021.05 | 25.2 | |
| Slow-8×8 (ours)FLOPs (G)=54.9, views=12021.05 | 24.1 | |
| Asyn-TF [37]backbone=VGG16, modality=RGB+Flow2018.12 | 22.4 | |
| Asyn-TFBackbone=VGG16, Feature bank=false, Pretrain=ImageNet2021.04 | 22.4 | |
| Asyn-TFFeatures=R + F2020.10 | 22.4 | |
| CoVIAR [54]backbone=R50, modality=Compressed2018.12 | 21.9 | |
| CoSlow_8FLOPs (G)=6.9, views=12021.05 | 21.5 | |
| Slow-8×8FLOPs (G)=54.9, views=12021.05 | 21.4 | |
| 2-Strm. [39]backbone=VGG16, modality=RGB+Flow2018.12 | 18.6 | |
| JMRNFeatures=P2020.10 | 16.2 | |
| I3DBackbone=R101-I3D, Modality=RGB2024.04 | 15.6 | |
| PA3DFeatures=P + CF + PAF2020.10 | 13.8 | |
| PoTionFeatures=P2020.10 | 13.54 | |
| 2StreamFeatures=R + F2020.10 | 11.9 |