Exercise Classification

Exercise classification research aims to automatically identify and categorize different types of physical activity, primarily to improve fitness tracking and personalized training. Current efforts focus on developing robust and efficient methods using various data sources, including wearable sensors (like IMUs) and video analysis, often employing deep learning architectures such as X3D and SlowFast networks, or leveraging techniques like sensor fusion and contrastive learning to enhance accuracy. These advancements hold significant promise for improving accessibility and personalization in fitness applications, particularly for individuals with disabilities or limited access to specialized equipment, and for advancing sports science research.