List your company

Spatio-temporal Action Localization on UCF101 24

88Video-mAP (IoU=0.2)STAR/L
39.32851.96464.677.236Nov 15, 2019Jun 11, 2020Jan 6, 2021Aug 4, 2021Mar 1, 2022Sep 26, 2022Apr 24, 2023

Evaluation results

MethodLinks
STAR/LPretraining=CLIP→K700, Backbone=ViViT/L2023.048890.3—71.835.2—
STAR/BPretraining=IN21K→K400, Backbone=ViViT/B2023.0487.787.3—66.230.9—
TubeR with flowPretraining=K400, Backbone=I3D2023.0485.381.3—60.229.7—
TubeRPretraining=IG65M→K400, Backbone=CSN-1522023.0483.383.2—58.428.9—
MOCPretraining=IN1K → COCO, Backbone=DLA342023.0482.878—53.828.3—
YOWO+LFBcausal=false, Long-term Feature Bank=true2019.1178.687.386.153.1——
ACT2019.1177.269.5—51.4——
ACTPretraining=IN1K, Backbone=VGG2023.0477.267.1—51.425—
STEP2019.1176.67583.1———
YOWO (16-frame)clip size=16-frame, causal=true2019.1175.880.482.548.8——
ROAD2019.1173.5——46.3——
MPS2019.1172.9—82.441.1——
Ours w/ Mix. Aug.Input=Skeleton2023.0361.8——38——
Anurag et al.Input=RGB2023.0361.7——35——
Ours w/o Mix. Aug.Input=Skeleton2023.0360.4——37.4——
T-CNN2019.1147.141.451.3———
Escorcia et al.Input=RGB2023.0345.5—————
Chéron et al.Input=RGB2023.0343.9——17.7——
Peng w/ MRMotion Region (MR)=with2019.1142.365.750.4———
Peng w/o MRMotion Region (MR)=without2019.1141.264.849.5———
Anurag et al.Input=RGB2026.04—————35
Cheron et al.Input=RGB2026.04—————17.7
SKPInput=Skeleton2026.04—————37.4
SKP + BatchMixInput=Skeleton2026.04—————38
SKP + Provenance-Based Input Gradient GuidanceInput=Skeleton2026.04—————39.7
UnifiedPretraining=K600, Backbone=SlowFast2023.04—79.3————