To learn about the health care of athletes in the knee joint, real-time to monitor the health of the human knee joint according to the cloud. The system uses a depth camera to collect the data information of the human lower limb alignment and obtains the spatial coordinate position of the human lower limb alignment through deep learning; then analyzes and processes the video sequence of the human lower limb alignment, including the wavelet function decomposition of the lower limb alignment information and reconstruction, and finally, the monitoring results were obtained by the knee joint scoring method. The system has completed the design of software and hardware and realized the method of extracting the coordinate information of the main joint points of the human body based on neural network. The health monitoring algorithm finally obtained the knee joint health status of the subjects through the evaluation system. By comparing the real health status of the subjects, the reliability of this research work was verified. The experiment shows that the monitoring error of the system is less than 10%, and the overall error is only 6%. The KSS standard score of the subjects is consistent with the monitoring and evaluation of the system, and the score trend is basically the same. For the situation that the overall score of the system is lower than the KSS standard, after communication and analysis with orthopaedic experts. It is speculated that it may be due to the subjective estimation of the subjects when measuring KSS, and the system monitoring algorithm is relatively strict and more objective. It is proved that the system designed in this paper can effectively monitor the health of athletes’ knee joint.