With the improvement of people's security awareness and the maturity of monitoring technology, the application field of intelligent monitoring system is expanding. The traditional monitoring system can realize the functions of video playback and storage. However, there exist some drawback such as the dependence on high-performance servers and purely manual monitoring mode, etc. Therefore, this paper constructs an intelligent monitoring system based on deep learning. Firstly, the system uses Jetson TX2 embedded platform to detect the skeleton points of video data collected from the camera, and achieves fall detection and face recognition through the human body posture recognition network. Then, the results are uploaded to the cloud server, processed by the cloud server and transmitted to the mobile App. Finally, once someone falls down, the mobile App will use voice and message push to alert users. Experiments show that the system can detect and alarm falls effectively.