A Literature Review: Geometric Methods and Their Applications in Human-Related Analysis

Area of Science:

• Computer Vision and Graphics
• Human-Computer Interaction
• Machine Learning

Background:

• Geometric features, including topological and manifold properties, are crucial for extracting geometric information.
• Geometric methods are widely applied in computer graphics and computer vision.
• Human-related analyses like shape, pose, and action analysis benefit significantly from geometric approaches.

Purpose of the Study:

• To provide a comprehensive literature review of geometric concepts, methods, and their applications in human-related analysis.
• To propose a categorization of geometric methods based on the scope of extracted properties.
• To investigate the emerging field of geometric deep learning and its relevance.

Main Methods:

• Categorization of geometric methods into object-oriented, feature-oriented, and routine-based approaches.
• Review of geometric deep learning techniques.
• Collection and comparison of validation datasets and method performances.

Main Results:

• A structured overview of geometric methods and their utility in human analysis.
• Identification of geometric deep learning as a significant research area.
• Comparative analysis of different methods based on performance metrics.

Conclusions:

• Geometric methods offer powerful tools for understanding human shape, pose, and actions.
• Geometric deep learning presents promising avenues for future research.
• Further research is needed to explore novel applications and refine existing geometric techniques.