Open Water Swimming Monitoring System Based on Multi-Task Learning

With the development of open water swimming, utilizing intelligent devices to assist swimmers in recording and analyzing their performance data, along with adding entertainment features, has emerged as a significant trend. However, there are some challenges such as difficulty in monitoring and the simplicity of functions. To address these issues, we propose a multifunctional open water swimming monitoring system based on visual technology. This system first achieves real-time visual monitoring and danger alerts for swimmers through a multitask monitoring device. It then employs a multitask algorithm to achieve swimming posture detection, boat recognition, and gesture recognition, with each task's detection accuracy exceeding 90%. Finally, through experimentation, our study has showcased the ability to analyze swimming postures, detect boat distances, and facilitate gesture-based human-machine interactions in open water swimming scenarios, showing excellent comprehensive performance and practical application value.