Simulation of infrared thermal images based on deep learning in athlete training: Simulation of thermal energy consumption

•The primary aim of this study is to present a groundbreaking image segmentation algorithm.•The research seeks to provide valuable tools for coaches and athletes alike.•Our algorithm improves the visualization of training effec. With the progress of science and technology and the development of sports science, the training methods of athletes are gradually developing towards a more scientific and data-oriented direction. Infrared thermal image technology can capture the temperature distribution of human body surface in real time. A large number of infrared thermal image data of athletes under different training intensity and environment are collected and input into the model for training. In the feature extraction stage, CNN can automatically identify the key temperature change region from the infrared thermal image. In the pattern recognition stage, the model can classify and predict the new thermal image data by learning the thermal energy consumption pattern under different training intensity. In the association learning stage, the model associates the thermal image features with the actual thermal energy consumption data, so as to achieve accurate simulation of thermal energy consumption. After a series of experiments and verification, the deep learning model constructed in this study shows high accuracy and reliability in the simulation of thermal energy consumption in infrared thermal images. The model can not only accurately identify the heat energy consumption pattern of athletes during training, but also predict the heat energy consumption that may occur under specific training conditions. The model also has good generalization ability, and can adapt to the heat energy consumption simulation needs of different athletes and different training environments.