Key result
An integrated deep learning model achieved 85.4% Top-1 accuracy for aerobic exercise recognition and found that aerobic exercise decreased college students' heart rates by 12.2%.
Why the study?
Existing action recognition models face challenges handling joint semantic information at different levels of granularity and indiscriminately processing all joints during training.
A novel deep learning model for aerobic exercise recognition achieved high accuracy and demonstrated physiological benefits of aerobic exercise in college students.
May enhance aerobic exercise recognition precision in students; leaves open clinical validation before any practice adoption.
Aerobic exercise awareness is a key factor in promoting the health and well-being of college students. However, existing action recognition models face significant challenges in handling joint semantic information at different levels of granularity and indiscriminately processing all joints during training. To address these issues, this work suggests a deep learning and neural network-based approach for recognizing it. To enable high-precision aerobic activity recognition, the system includes a multi-granularity spatio-temporal graph convolutional network (MGST-GCN), a long short-term memory (LSTM) network, and a convolutional neural network (CNN). In the research, the integrated model on the NTU-RGB+D dataset obtains a Top-1 accuracy of 85.4% and a Top-5 accuracy of 95.2%. Furthermore, using the Kinetics-Skeleton dataset, this model obtains a Top-1 accuracy of 87.6% and a Top-5 accuracy of 96.8%. The comparative models (3D-CNN and two-stream CNN) are substantially underperformed by these outcomes. This investigation assesses how aerobic exercise affects college students’ health by using the integrated model. College students’ heart rates decrease by 12.2%, their exercise intensity increases by 13.3%, their exercise duration increases by 15[Formula: see text]min, their calorie consumption increases by 20.8%, and their frequency of exercise increases by 33.3% when they engage in aerobic exercise, such as running, according to the experimental results. These findings indicate that this work not only makes an innovative contribution to the field of action recognition but also provides effective support for health management among college students. Future research will pay more attention to enhancing the model’s generalization ability, expanding the coverage of datasets, and introducing more health indicators to further enhance the system’s performance and application scope.
No takes yet. Share an insight, caveat, or question.
A 2025 study studied Aerobic exercise recognition for college students' health. Integrated deep learning model (MGST-GCN, LSTM, CNN) vs. 3D-CNN and two-stream CNN was evaluated on Top-1 accuracy on the NTU-RGB+D dataset. An integrated deep learning model achieved 85.4% Top-1 accuracy for aerobic exercise recognition and found that aerobic exercise decreased college students' heart rates by 12.2%.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: