HandMCM: Multi-modal Point Cloud-based Correspondence State Space Model for 3D Hand Pose Estimation
This work addresses occlusion issues in 3D hand pose estimation for human-computer interaction applications like augmented reality, representing an incremental improvement with strong specific gains.
The paper tackled 3D hand pose estimation under occlusion challenges by proposing HandMCM, a multi-modal point cloud-based correspondence state space model, which significantly outperformed state-of-the-art methods on benchmark datasets, especially in severe occlusion scenarios.
3D hand pose estimation that involves accurate estimation of 3D human hand keypoint locations is crucial for many human-computer interaction applications such as augmented reality. However, this task poses significant challenges due to self-occlusion of the hands and occlusions caused by interactions with objects. In this paper, we propose HandMCM to address these challenges. Our HandMCM is a novel method based on the powerful state space model (Mamba). By incorporating modules for local information injection/filtering and correspondence modeling, the proposed correspondence Mamba effectively learns the highly dynamic kinematic topology of keypoints across various occlusion scenarios. Moreover, by integrating multi-modal image features, we enhance the robustness and representational capacity of the input, leading to more accurate hand pose estimation. Empirical evaluations on three benchmark datasets demonstrate that our model significantly outperforms current state-of-the-art methods, particularly in challenging scenarios involving severe occlusions. These results highlight the potential of our approach to advance the accuracy and reliability of 3D hand pose estimation in practical applications.