YOLO-CBD: A Recognition and Detection Method for Cigarette Box Based on YOLOv5

Abstract

In recent years, cigarette box recognition has be-come a significant aspect of object detection that has applications in many areas, such as the retail industry and health care. This paper proposes a real-time YOLOv5-based cigarette box detection method (YOLO-CBD). Specifically, the approach involves training YOLOv5 to detect and label the location of cigarette boxes in an image, then using the ArcFace algorithm to extract the feature of cigarette boxes. We evaluate the YOLO-CBD method on the cigarette box datasets. The results show that our method achieves a high MAP of 95.1% in complex scenarios, which is better than the state-of-the-art methods. In conclusion, the YOLO-CBD demonstrates high accuracy and stability in detecting cigarette boxes. It is a promising approach for practical applications in the tobacco industry.