Year: 2026
Type: preprint
Abstract: In recent years, as the impact of deep learning on people's lives has grown, more and more attention has been paid to the development of this field. The task of human hand pose and shape estimation from an RGB image has been a long-standing problem in the field of computer vision. Unlike common hand posture prediction, which only predicts the coordinates of the skeletal points of the hand, this task restores the original shape of the hand. Many places will apply this task such as augmented reality and virtual reality, but the task is still very challenging because the hand occupies a relatively small part of the image area, and the hand movements are flexible and easy to block. In this paper, we propose an end-to-end network for 3D hand mesh reconstruction from RGB images. Using ResNet-50 as the encoder, we extract image features and generate 2D heatmaps and masks through convolutional layers to improve parameter regression. The model parameters are iteratively refined with fully connected layers. To correct unnatural hand poses, we apply a graph convolutional network to adjust mesh coordinates, combining these adjustments with the initial MANO-generated model. Experiments demonstrate effective 3D hand mesh reconstruction from monocular images.
Source: Research Square
Authors: Tsung-Han Tsai, Jia-Yang Jiang
Institution: National Central University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7235140686
