In recent years, building collapse accidents have occurred from time to time in China, causing serious casualties and property losses. In order to monitor the abnormal phenomena of buildings in time, Intelligent Building Safety Monitoring System based on computer vision is proposed. Using ZigBee wireless sensor network, the crack images and sensor monitoring data of monitoring points are obtained and transmitted to the local control computer. After the data are processed by the integrated crack image recognition and measurement algorithm based on OpenCV, they are displayed on the terminal monitoring software. If the data is abnormal, the early warning function is triggered, and the intelligent monitoring and danger early warning functions for building safety are realized.