Home News Latest Papers Tutorials Datasets Wiki SOTA LLM Models GPU Leaderboard Events

English

Action Recognition On H2O 2 Hands And Objects

Metrics

Actions Top-1

Hand Pose

Object Label

Object Pose

RGB

Results

Performance results of various models on this benchmark

Model Name	Actions Top-1	Hand Pose	Object Label	Object Pose	RGB	Paper Title	Repository
RGBPoseConv3D	83.47	2D	No	No	Yes	Revisiting Skeleton-based Action Recognition
MS-G3D	50.83	3D	No	No	No	Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition
SHARP	91.73	3D	Yes	2D	No	SHARP: Segmentation of Hands and Arms by Range using Pseudo-Depth for Enhanced Egocentric 3D Hand Pose Estimation and Action Recognition
HTT	86.36	3D (est.)	Yes (est.)	No	Yes	Hierarchical Temporal Transformer for 3D Hand Pose Estimation and Action Recognition from Egocentric RGB Videos
ISTA-Net	89.09	3D	No	Yes	No	Interactive Spatiotemporal Token Attention Network for Skeleton-based General Interactive Action Recognition
TA-GCN	79.25	3D	No	Yes	No	H2O: Two Hands Manipulating Objects for First Person Interaction Recognition	-
H2OTR	90.90	3D (est.)	Yes (est.)	Yes (est.)	Yes	Transformer-Based Unified Recognition of Two Hands Manipulating Objects
SlowFast	77.69	No	No	No	Yes	SlowFast Networks for Video Recognition
ST-GCN	73.86	3D	No	Yes	No	Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition
EffHandEgoNet	91.32	2D	Yes	2D	No	In My Perspective, In My Hands: Accurate Egocentric 2D Hand Pose and Action Recognition
HandFormer-B/21x8	93.39	3D	No	No	Yes	On the Utility of 3D Hand Poses for Action Recognition

0 of 11 row(s) selected.