Express Carton Detection Based On Improved YOLOX

The high-quality development of the logistics industry is an important driving force for the Chinese economy's high-quality development. This paper proposes an improved YOLOX model based on existing target detection work to solve problems in the traditional logistics industry's transit, sorting, management, and other aspects of express delivery. And this paper employs transfer learning to compensate for a lack of model training data. In order to improve YOLOX's detection performance for express cartons, three attention methods (SENet, CBAM, and ECA) are introduced and compared in this work, and the experimental results show the feasibility and effectiveness of improved YOLOX model on the express carton dataset. And the ECA-YOLOX model outperforms the other three models (YOLOX, SENet-YOLOX, and CBAM-YOLOX) and has the greatest metrics for precision, recall, F1 score, and mAP.