<![CDATA[Accurate and efficient detection of tobacco quality is essential to identify the quality of tobacco leaves and improve farmers' tobacco yields. However, due to the high similarity between classes, significant intraclass differences and complex backgrounds among different tobacco leaves, accurately identifying tobacco quality through neural network models can pose significant challenges. To address this problem, this paper is presented with a fast and accurate method of detecting and identifying tobacco leaf quality using YOLO (You Only Look Once). This model uses YOLOv11 which incorporates an efficient detection head designed to detect the characteristics of tobacco leaves. In addition, an in-depth surveillance layer is introduced into the network along with incorporating and enhancing dynamic upsampling modules. Experimental data include public data sets of L1L, L2L, L3L, L3R, L4R, L10, L10F, and LND tobacco leaf quality. The results of the experiment showed that Yolov11 outperformed the Yolov8 and Yolov12 algorithms with a precision of 0.853, an F1 score of 0.73 mAP@0.5, 0.811 and mAP@0.5:0.95 of 0.631.]]>
Copyrights © 2026
