In order to accurately assess the health status of athletes’ legs and recognize athletes’ leg injuries in a timely manner, a study on the identification of athletes’ leg injuries in tennis training was carried out in the context of IoT. First, the athletes were equipped with wearable Internet of Things (IoT) devices to monitor and record the physiological parameters and movement data of athletes in real time during their training process, and to collect information about sports injuries. Second, images of athletes’ sports injuries from tennis training were processed to extract the characteristics of leg injuries. Based on this, an injury recognition model was constructed using a support vector machine to comprehensively recognize leg injuries in tennis training. Experimental data validation demonstrates that this method can accurately identify all potentially damaged joints in athletes, ensuring no misjudgments or omissions.