EgoEV-HandPose: Egocentric 3D Hand Pose Estimation and Gesture Recognition with Stereo Event Cameras
EgoEV-HandPose: Egocentric 3D Hand Pose Estimation and Gesture Recognition with Stereo Event Cameras
This paper introduces EgoEV-HandPose, a framework for joint 3D bimanual hand pose estimation and gesture recognition using stereo event streams. The core KeypointBEV module combines sparse 3D joint hypotheses with geometry-aware stereo fusion and iterative reprojection-guided refinement to resolve depth uncertainty. The authors also release EgoEVHands, the first large-scale real-world stereo event dataset for this task, containing 2,498 annotated stereo sequence pairs across 38 gesture classes. Experiments show a mean per-joint position error (MPJPE) of 30.54mm and gesture recognition accuracy of 86.87%, outperforming RGB-based stereo and prior event-based methods, with demonstrated robustness under low illumination and severe occlusion.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
trajectories
2,498
Trajectory count
View original evidence
real-world stereo event dataset for egocentric 3D bimanual hand pose estimation and gesture recognition. Built upon EgoEvGesture recordings, it adds dense 2D/3D hand annotations and comprises 2,498 annotated stereo sequence pairs across 38 gesture classes. Extensive experiments demonstrate that EgoEV-HandPose achieves state-of-the-art performance on EgoEVHands, obtaining a mean per-joint position error (MPJPE) of 30
Open source S4percent
86.87
View original evidence
.54mm and a gesture recognition accuracy of 86.87%, significantly outperforming RGB-based stereo and previous event-based methods. The framework exhibits exceptional robustness under low illumination and severe bimanual occlusion. Experiments on DHP19 further demonstrate the cross-dataset and cross-architecture generalizability of the KeypointBEV refinement design. Dataset and code will be released at https://github.
Open source S5
- paper: not_reported
- code: Reported pending Open source S5
- weights: not_reported
- dataset: Reported pending Open source S5
- code: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
atasets. We propose EgoEV-HandPose, an end-to-end framework for joint 3D bimanual hand pose estimation and gesture recognition from stereo event streams. Its core module, KeypointBEV, combines sparse 3D joint hypotheses with geometry-aware stereo fusion and iterative reprojection-guided refinement to progressively resolve depth uncertainty and improve pose accuracy. We also introduce EgoEVHands, the first large-scale
Open source S3
real-world stereo event dataset for egocentric 3D bimanual hand pose estimation and gesture recognition. Built upon EgoEvGesture recordings, it adds dense 2D/3D hand annotations and comprises 2,498 annotated stereo sequence pairs across 38 gesture classes. Extensive experiments demonstrate that EgoEV-HandPose achieves state-of-the-art performance on EgoEVHands, obtaining a mean per-joint position error (MPJPE) of 30
Open source S4
.54mm and a gesture recognition accuracy of 86.87%, significantly outperforming RGB-based stereo and previous event-based methods. The framework exhibits exceptional robustness under low illumination and severe bimanual occlusion. Experiments on DHP19 further demonstrate the cross-dataset and cross-architecture generalizability of the KeypointBEV refinement design. Dataset and code will be released at https://github.
Open source S5
EgoEV-HandPose: Egocentric 3D Hand Pose Estimation and Gesture Recognition with Stereo Event Cameras
Open source S1
arXiv:2605.12297v2 Announce Type: replace-cross Abstract: Egocentric 3D hand pose estimation and gesture recognition are fundamental to understanding hand interactions from first-person observations. Robust perception remains challenging: frame-based cameras suffer from motion blur and limited dynamic range, while event-based approaches face ego-motion interference, monocular depth ambiguity, and limited real-world d
Open source S2
Source：arXiv Robotics — research abstracts · arxiv.org