Abstract: Aiming at the problem that readers are not familiar with the arrangement method of call numbers, which makes it inconvenient to find books and takes a long time, a call number recognition algorithm for library books based on improved YOLOv5s and OCR technology is proposed. The algorithm consists of two parts. The first is the improved YOLOv5 algorithm to identify the call number label area of the spine of the book. The second is that EasyOCR recognizes the call number text in the area. In order to improve the detection speed, the YOLOv5s algorithm is lightweight improved. Firstly, the lightweight MobileNetV3 is used as the backbone network to reduce the amount of parameters and calculation. Secondly, the more efficient SimSPPF is used to improve the original SPP to speed up the network operation. Finally, the CA coordinate attention mechanism is embedded in the feature fusion network to improve the recognition accuracy of the model. The experimental results show that the proposed algorithm has higher recognition accuracy and faster speed than the original algorithm, and can quickly and accurately identify the book call number. It is helpful to promote the development of libraries from digital to intelligent and provide readers with more convenient, efficient and personalized services.