Action Recognition In Videos On Ntu Rgbd

评估指标

Accuracy (CS)

Accuracy (CV)

评测结果

各个模型在此基准测试上的表现结果

			Paper Title	Repository
DSCNet (RGB + Pose)	97.4	99.4	A Dense-Sparse Complementary Network for Human Action Recognition based on RGB and Skeleton Modalities	-
PoseC3D (RGB + Pose)	97.0	99.6	Revisiting Skeleton-based Action Recognition
π-ViT (RGB + Pose)	96.3	99.0	Just Add $\pi$! Pose Induced Video Transformers for Understanding Activities of Daily Living
UMDR (RGB-D)	96.2	98.0	A Unified Multimodal De- and Re-coupling Framework for RGB-D Motion Recognition
MMNet (RGB + Pose)	96.0	98.8	MMNet: A Model-Based Multimodal Network for Human Action Recognition in RGB-D Videos	-
Hierarchical Action Classification (RGB + Pose)	95.66	98.79	Hierarchical Action Classification with Network Pruning	-
VPN (RGB + Pose)	95.5	98.0	VPN: Learning Video-Pose Embedding for Activities of Daily Living
EPP-Net (Parsing + Pose)	94.7	97.7	Explore Human Parsing Modality for Action Recognition
3DA (RGB + Pose)	94.3	97.9	Cross-Modal Learning with 3D Deformable Attention for Action Recognition	-
Action Machine (RGB only)	94.3	97.2	Action Machine: Rethinking Action Recognition in Trimmed Videos	-
π-ViT (RGB only)	94.0	97.9	Just Add $\pi$! Pose Induced Video Transformers for Understanding Activities of Daily Living
IPP-Net (Parsing + Pose)	93.8	97.1	Integrating Human Parsing and Pose Network for Human Action Recognition
ViewCon (RGB + Pose)	93.7	98.9	Multi-View Action Recognition Using Contrastive Learning	-
DVANet (RGB only)	93.4	98.1	DVANet: Disentangling View and Action Features for Multi-View Action Recognition
TSMF (RGB + Pose)	92.5	97.4	Multimodal Fusion via Teacher-Student Network for Indoor Action Recognition	-
MSAF (RGB+Pose)	92.24	-	MSAF: Multimodal Split Attention Fusion
STAR-Transformer (RGB + Pose)	92.0	96.5	STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition	-
MMTM (RGB+Pose)	91.99	-	MMTM: Multimodal Transfer Module for CNN Fusion
FUSION (IR+Pose)	91.8	94.9	Infrared and 3D skeleton feature fusion for RGB-D action recognition
PoseMap (RGB+Pose)	91.7	95.2	Recognizing Human Actions as the Evolution of Pose Estimation Maps	-

0 of 25 row(s) selected.