Every exam week, university libraries will be in short supply and maliciously occupy seats. In order to solve this situation, we designed the library as the detection system based on image recognition and deep learning. The detection system includes three parts: video intelligent processing end, server back-end data storage section and wechat applet display end. At the video intelligent processing end, we realize the recognition of seat usage in the image based on OpenCV image processing and deep learning vgg16 algorithm; At the back-end data storage end of the server, on the one hand, it can return to the wechat applet to show the usage of each seat, on the other hand, it can update the content of the database according to the data transmitted from the video processing end; On the wechat applet display side, we showed users the use of Library seats. The library seat detection system can not only make full use of the library seat resources to a certain extent, but also reduce the working pressure of the library staff.