Answers are grounded in this paper and cite their sources.
Evaluates injury recovery effects in college students, indicating enhanced health tracking capabilities.
To address the challenges of high-dynamic motion analysis and health tracking in smart gymnasiums, this paper proposes the Student Pose-aware Recovery Network (STUPOSRENet). Taking multi-angle high-definition physical education video streams as input, the model synchronously captures spatial morphology and temporal trajectories to achieve high-precision identification of movement postures and parallel quantitative evaluation of injury recovery effects. The network ultimately outputs multi-task analytical results, explicitly defined as discrete motion action classification labels, Gaussian-smoothed optimized skeletal joint coordinates, and percentage-based quantitative knee joint injury recovery efficiency scores. This model introduces a multi-scale spatial attention mechanism and a special health assessment branch, which can resist dynamic occlusion and complex background interference. The experimental results on the HPU-Sports dataset show that the model has reached 95.2% accuracy in long jump posture recognition, which is 7.8% higher than AlphaPose in complex hurdle take-off forms. In the 12-week rehabilitation monitoring experiment, the recovery efficiency score of the injured parts increased from 45% in the first week to 95%, showing a high sensitivity in tracking kinematic rehabilitation trends. The inference latency of the system is only 22 milliseconds, and the comprehensive efficiency index is 7.5, which is about 4.1 times that of OpenPose, confirming that its real-time deployment in smart sports classrooms has high efficiency and engineering value.
No takes yet. Share an insight, caveat, or question.
Ren et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: