HyperAIHyperAI초신경
홈뉴스최신 연구 논문튜토리얼데이터셋백과사전SOTALLM 모델GPU 랭킹컨퍼런스
전체 검색
소개
HyperAIHyperAI초신경

Temporal Action Localization On Fineaction

평가 지표

mAP
mAP [email protected]
mAP [email protected]
mAP [email protected]

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름
mAP
mAP [email protected]
mAP [email protected]
mAP [email protected]
Paper TitleRepository
DBG (i3d feature)6.7510.656.432.50Fast Learning of Temporal Action Proposal via Dense Boundary Generator-
ActionMamba(InternVideo2-6B)29.0445.4428.826.79Video Mamba Suite: State Space Model as a Versatile Alternative for Video Understanding-
DyFADet (VideoMAE v2-g)23.837.123.75.9DyFADet: Dynamic Feature Aggregation for Temporal Action Detection-
RDFA-S6 (InternVideo2-6B)29.646.429.57.6Enhancing Temporal Action Localization: Advanced S6 Modeling with Recurrent Mechanism-
G-TAD (i3d feature)9.0613.748.833.06G-TAD: Sub-Graph Localization for Temporal Action Detection-
BMN (i3d feaure)9.2514.448.923.12BMN: Boundary-Matching Network for Temporal Action Proposal Generation-
InternVideo17.57---InternVideo: General Video Foundation Models via Generative and Discriminative Learning-
InternVideo2-6B27.7---InternVideo2: Scaling Foundation Models for Multimodal Video Understanding-
VideoMAE V2-g18.2429.0717.665.07VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking-
0 of 9 row(s) selected.
Temporal Action Localization On Fineaction | SOTA | HyperAI초신경