Tobacco shred varieties detection based on Yolov7

The composition distribution detection of processed tobaccos is one of the main difficulties in cigarette design and processing. The radio of the four kinds of tobacco influences the quality of cigarettes. Therefore, it is very important to detect tobacco shreds on the tobacco processing in order to ensure the cigarette consistency of formulation. Aiming at solving the problems of poor efficiency and low accuracy of traditional method and utilizing the advantages of deep leaning network in object detection, a method based on Yolov7 for tobacco shred detection is proposed. The collected images of four categorized tobacco shreds are manually marked, and model is trained through Yolov7 network. The mean average precision of tobacco shred detection in this work reached 0.986, and the tobacco shred detection speed was also vastly increased compared to traditional manual detection method.