March 2026
Deep Computer Vision and Transfer Learning Approach for Multi-Class Mango Leaf Disease Detection and Segmentation
Developed a YOLOv12-based computer vision framework for multi-class mango leaf disease detection and instance segmentation using transfer learning. Trained on 3,479 annotated images across seven healthy and diseased leaf classes, the model achieved 1.00 mAP@0.5 for detection, 0.99 mAP@0.5 for segmentation, and 100% recall across disease categories. The results demonstrate strong robustness and generalization under varying lighting, orientations, and disease severity, supporting applications in precision agriculture and automated crop monitoring.