To strengthen the safety management of construction site and intelligently detect the helmets worn by workers,this paper proposes an intelligent video monitoring system for navigation and power project based on object detection,which consists of object detection module and network communication module,deployed according to the C/S architecture,and successfully applied to the construction site of navigation and power hub project on Minjiang River.The object detection module uses advanced object detection algorithm to extract,classify and locate the features of safety belts and helmets in the image through convolutional neural network,and thus complete the recognition of safety belts and helmets.The network communication module connects the video capture terminal,server and video player terminal through the network to complete the image capture,transmission and display functions.The test of each component of the system shows that the system has complete functions and the modules are well connected.It can complete the real-time image acquisition,transmission,detection and display functions of each scene in the process of project construction,providing a solid foundation for the safety management of Longxikou navigation and power project construction.