Express Sorting System Based on Improved YOLO Algorithm

Abstract

With the rapid development of e-commerce, the volume of parcel handling in the logistics industry has increased dramatically, and traditional manual sorting faces the problems of low efficiency and high cost. Based on the above problems, this paper develops an express sorting robot system based on computer vision. The system selects Raspberry Pi as the main control module and utilizes a variety of algorithms, among which the target detection adopts the YOLO series of algorithms. Meanwhile, this paper proposes a new courier sorting dataset, Parcel360° Dataset, which is utilized to train and evaluate YOLOv5, YOLOv8 and YOLO11. Experiments show that YOLOv8 achieves the best balance between accuracy and efficiency, with a mean accuracy (mAP) of 62.8% and a frame rate (fps) of 27, which can meet the high accuracy and real-time requirements of logistics sorting. The system helps to promote the intelligent transformation of the logistics industry and enhance the overall competitiveness of the industry.