A real-time correction system for track and field sprinting posture based on machine vision feedback
Lingling Zhang · Discover Computing · 2026
To address the issues of low movement efficiency and increased injury risk caused by improper running posture in track and field short-distance running, a real-time running posture correction system based on machine vision feedback is proposed. This system is centered on machine vision technology, using multiple cameras to simultaneously capture running motion images. After preprocessing, posture feature extraction, and anomaly identification, targeted correction instructions are generated and immediately fed back to the athletes. A four-module topological architecture is designed, integrating image acquisition, feature calculation, intelligent recognition, and feedback control units, to optimize the posture recognition algorithm and correction strategy. Through the construction of an experimental platform, 20 short-distance runners were selected for testing to verify the real-time performance, accuracy, and correction effect of the system. The experimental
Results: show that under controlled indoor conditions, the accuracy of posture feature recognition reaches 96.3% (95%CI: 95.8%-96.8%), the correction response delay is ≤ 100ms, and it can effectively optimize the key sprint posture indicators including stride uniformity, arm swing angle, and trunk pitch angle of the athletes; the 100 m sprint performance of the subjects in the experimental group was improved by an average of 2.87% ( p < 0.05), and the injury risk assessment value based on
Objective: biomechanical indicators was reduced by an average of 31.2% ( p < 0.01) in the experimental scenario. Mediation effect analysis verified that the optimized biomechanical load mediated 78.4% of the injury risk reduction effect and 82.1% of the performance improvement effect, supporting a plausible causal pathway of posture correction. This research provides technical support for the scientific training of short-distance running and has strong engineering application value and academic reference
Significance: .