Aiming at the problem of missing assembly that may easily occurs in the assembly process of mechanical products,an assembly process monitoring system based on deep learning was proposed with the RV reducer as the research object.An assembly process monitoring system experiment platform was built,the depth images and color images of assembly were collected using RGBD cameras.The semantic segmentation and target detection algorithm based on deep learning was used to predict the collected depth images and color images to obtain prediction results.Functions such as deep learning algorithm and video capture were integrated into the assembly process monitoring software to realize the monitoring of RV reducer assembly process.The experiments have proved that:the software can correctly monitor whether each part is missing in the assembly process,and has a certain real-time performance.