Physical exercises are important for enhancing the quality of life of the trainer. However, most exercises just include simple movements repeated during training. This characteristic could reduce the trainer’s motivation during long-term training, especially in the case of rehabilitation exercises. Serious games can enhance cognitive effects for training, but most serious game systems must be facilitated with complex sensor systems for capturing human motions. This drawback can cause difficulties for the trainers. Consequently, in this study, we designed a prototype of a contactless sensor-based serious game system to remotivate exercise trainers. We first focused on the squat game, which is one of the most popular lower limb exercises. To simplify the system, we employed only a mono camera to track the body skeleton in real-time. The center point of the skeleton was used for controlling the object of the game system. In the game system, we designed a meta object that can be moved up and down to avoid collision with obstacles. The obstacles were positioned so that the user could stand up and down like doing a squat to pass them. The trainers will be scored according to the number of obstacles being passed. As a result, during the game, the trainers will focus their attention on getting as high as possible rather than training the squat. In perspective, we will develop the system for other exercises and implement musculoskeletal simulation for analyzing muscles during the training. The further system could also be employed for rehabilitation exercises.