Action Classification On Moments In Time

評価指標

Top 1 Accuracy

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名	Top 1 Accuracy	Paper Title	Repository
MoViNet-A5	39.1	MoViNets: Mobile Video Networks for Efficient Video Recognition
MoViNet-A4	37.9	MoViNets: Mobile Video Networks for Efficient Video Recognition
UMT-L (ViT-L/16)	48.7	Unmasked Teacher: Towards Training-Efficient Video Foundation Models
I3D	29.51%	Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
TSN-2Stream	-	Temporal Segment Networks for Action Recognition in Videos
SRTG r3d-34	28.55	Learn to cycle: Time-consistent feature discovery for action recognition
MoViNet-A0	27.5	MoViNets: Mobile Video Networks for Efficient Video Recognition
UniFormerV2-L	47.8	UniFormerV2: Spatiotemporal Learning by Arming Image ViTs with Video UniFormer
TRN-Multiscale	28.27	Temporal Relational Reasoning in Videos
SRTG r3d-101	33.56	Learn to cycle: Time-consistent feature discovery for action recognition
AssembleNet	34.27%	AssembleNet: Searching for Multi-Stream Neural Connectivity in Video Architectures
SRTG r3d-50	30.72	Learn to cycle: Time-consistent feature discovery for action recognition
CoVeR(JFT-3B)	46.1	Co-training Transformer with Videos and Images Improves Action Recognition	-
EvaNet	31.8%	Evolving Space-Time Neural Architectures for Videos	-
InternVideo2-1B	50.9	InternVideo2: Scaling Foundation Models for Multimodal Video Understanding
CoVeR(JFT-300M)	45.0	Co-training Transformer with Videos and Images Improves Action Recognition	-
CoST (ResNet-101, 32 frames)	32.4%	Collaborative Spatiotemporal Feature Learning for Video Action Recognition
OmniVec2	53.1	OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning	-
MBT (AV)	37.3	Attention Bottlenecks for Multimodal Fusion
ViViT-L/16x2	-	ViViT: A Video Vision Transformer

0 of 29 row(s) selected.