Related Experiment Video
Updated: Jun 3, 2026
09:41
Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
Published on: April 21, 2023
Spatial-Temporal Self-Compensating Graph Convolutional Network for Skeleton-Based Action Recognition Under Data
Summary
This study introduces a Spatial Temporal Self-compensating Graph Convolutional Network (STSc-GCN) to improve skeleton-based human action recognition. The novel network enhances robustness against real-world data constraints like occlusion and missing frames.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Machine Learning
Background:
- Skeleton-based human action recognition is crucial for computer vision.
- Existing methods struggle with real-world data issues like occlusion and missing frames.
- These limitations reduce the practical applicability of current techniques.
Purpose of the Study:
- To develop a robust method for skeleton-based human action recognition.
- To address performance degradation caused by data constraints.
- To improve the reliability of action recognition in real-world scenarios.
Main Methods:
- Propose a Spatial Temporal Self-compensating Graph Convolutional Network (STSc-GCN).
- Utilize a data self-compensation mechanism leveraging human movement patterns.
- Implement Collaborative Motion Spatial Compensation (CMSC) and Meta-action Sharpening Temporal Compensation (MSTC) modules.
Main Results:
- STSc-GCN achieves state-of-the-art performance on four constrained datasets.
- Demonstrates superior results on three standard datasets.
- Confirms effectiveness in both constrained and general scenarios.
Conclusions:
- STSc-GCN effectively mitigates performance degradation from data constraints.
- The proposed method enhances robustness and adaptability in human action recognition.
- The approach shows significant promise for real-world computer vision applications.