Hand grasping posture data enhancement method and system based on contact graph
By generating a data-enhanced hand grasping posture model using contact graph technology, the problem of error distribution between hand posture and model posture was solved. This achieved increased data volume and improved generalization ability, ensuring the matching of hand posture and the generation effect.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-03-24
AI Technical Summary
Existing hand grasping posture generation models suffer from errors in hand posture distribution compared to the model's posture distribution during data augmentation due to traditional methods. This results in poor model generalization ability and generation performance, especially when dealing with out-of-dataset models that generate unreasonable postures.
By using a contact map-based hand grasping posture data augmentation method, a sampled dataset of the target model is obtained, a contact map is generated, and a data-augmented contact model is designed using contact map technology. This includes data processing of point cloud or mesh types, performing expansion operations and geometry generation in the non-contact domain to ensure hand posture matching.
It effectively enhances hand pose data, increases the amount of data, improves the generalization and inference capabilities of neural networks, avoids the pose mismatch problem in traditional methods, and improves the generalization ability and generation effect of the model.
Smart Images

Figure CN121725152A_ABST