Research on Tobacco Foreign Object Detection Based on Deep Learning of Texture Features
Di Wu, Liang Chen, Qian Gu, Luo Ji · 2023
The content of foreign objects in tobacco directly affects the sensory quality of cigarette products. As tobacco leaves are agricultural products, it is inevitable that some foreign objects are mixed in during the picking, transportation, and other processes. In the process of cigarette manufacturing, We need to detect foreign objects from the fast-moving production line in time to prevent them from flowing into the next process. Considering the real-time property of detection and the unique texture characteristics of tobacco leaves, this paper proposes a tobacco foreign object detection method based on texture features and YOLO v8 model, using single-channel grayscale images. The experimental results show that the mAP–0.5 of the algorithm in this paper is 98.5%; when the confidence is 0.967, the detection precision is 100%. The average prediction time for each image is 27.1 ms, which meets the real-time requirement. This proposed scheme has the advantages of fast foreign object detection speed, low cost, and easy deployment, and has good promotional value.