List your company

Video Action Recognition on SS v2 (ACC@1)

76.8Top-1 Accuracy (SS v2)VideoMAE V2
-0.78419.35839.559.642Sep 25, 2025Nov 1, 2025Dec 8, 2025Jan 15, 2026Feb 21, 2026Mar 30, 2026May 7, 2026

Evaluation results

MethodLinks
VideoMAE V22025.0976.8
V-JEPA2-LModel Size=Large2026.0373.7
V-JEPA-2Params (M)=355, FLOPs (G×T×S)=935×2×32026.0573.7
LookWhen (70% sparse)Params (M)=106, FLOPs (G×T×S)=108×2×3, sparsity=70%2026.0572
VideoMAEv2Params (M)=87, FLOPs (G×T×S)=180×2×32026.0571.2
VideoMamba-M800eParams (M)=74, FLOPs (G×T×S)=202×2×32026.0571
UMT-B800eParams (M)=87, FLOPs (G×T×S)=180×2×32026.0570.8
VideoMAEParams (M)=87, FLOPs (G×T×S)=180×2×32026.0570.8
VideoMAE + RLT (τ=0.1)Params (M)=87, FLOPs (G×T×S)=120×4×32026.0570.2
InternVideo2Params (M)=1020, FLOPs (G×T×S)=2500×2×32026.0569.7
VideoMAE + ToMe (r=64)Params (M)=87, FLOPs (G×T×S)=131×4×32026.0569.7
VideoMAE + vid-TLDRParams (M)=87, FLOPs (G×T×S)=57×unk2026.0569.6
VideoMambaProParams (M)=72, FLOPs (G×T×S)=183×4×32026.0569.4
LookWhen (90% sparse)Params (M)=106, FLOPs (G×T×S)=40×2×3, sparsity=90%2026.0569.3
OmniStream2026.0368.5
VideoMAE + LITE (K=0.3)Params (M)=87, FLOPs (G×T×S)=46×2×32026.0568.3
VideoSwin + STTSParams (M)=89, FLOPs (G×T×S)=190×1×32026.0568.1
NoFrame2025.0962.7
TSM2025.0961.7
DINOv3-LModel Size=Large2026.0354
MoTIF-STBackbone=PE-G/142025.0941.9
MoTIFBackbone=PE-G/142025.0940.4
MoTIFBackbone=PE-L/142025.0937.3
Global CBMBackbone=PE-G/14, Evaluation Protocol=Supervised2025.0933.6
MoTIFBackbone=ViT-L/142025.0925.8
Zero-shotBackbone=PE-G/14, Evaluation Protocol=Zero-shot2025.092.2