LIGHTWEIGHT ARTIFICIAL INTELLIGENCE FOR PHYSICAL ACTIVITY RECOGNITION BASED ON WEARABLE DEVICES: A SYSTEMATIC LITERATURE REVIEW

Abstract

With the development of artificial intelligence, physical activity recognition has begun to play an important role for healthcare, sports, safety, and improving user interaction in intelligent environments. The active proliferation of wearable devices with multimodal sensors provides data collection, which can be used to improve the quality of life. Conventional AI models require large computing resources, so they are not suitable for devices with low performance. The study of lightweight AI, combined with the optimization of machine learning models, is gaining relevance. Such models work in conditions of limited memory and energy resources. Given the rapid technological progress and the latest relevant research, it is important to examine the current state of the physical activity recognition system, highlighting its strengths, as well as the challenges that this area continues to face. This study provides a systematic review of the literature on lightweight artificial intelligence for recognizing physical activity based on wearable devices, based on an analysis of articles published from 2010 to 2025.