Adaptive rescaling technique for portable vision devices in IoMT toward swimming workouts training and safety

In sports, intelligent multimedia models deployed through edge–cloud networks are widely used to build Internet of Things (IoT) capabilities. In amateur sports, however, the rapid use of personal devices, cameras, and sensors has made implementation problems increasingly significant, particularly for athlete health monitoring and basic sports training. In swimming training and amateur workouts, there is a growing need for unsupervised findings with tiny artificial intelligence (TinyAI) algorithms that can enhance image quality and perform content classification, enabling trainers to quickly and effectively provide on-site feedback to swimmers. In this paper, we propose a novel algorithm capable of improving the quality of rescaled images while preserving visual edges in swimming pool environments under various conditions, such as wavy water surfaces. The proposed method is lightweight and can be rapidly deployed on commercial sports IoT devices at the network edge. Experimental comparisons with existing approaches demonstrate that our method outperforms state-of-the-art techniques in terms of both quantitative performance and visual edge preservation.