Plant Disease Detection and Classification with Deep Learning

— In response to the significant losses caused by plant diseases, which call for a 70% increase in global food supply by 2050, scientists have created sophisticated deep learning models. These models, which were originally trained on datasets such as PlantVillage, have difficulties in real-world settings because of the intricate backgrounds and numerous leaves in each picture. This work presents FieldPlant, a new dataset with 5,170 photos of plant diseases that were gathered from plantations and painstakingly annotated by plant pathologists. The research assesses state-of-the-art classification algorithms such as MobileNet, VGG16, InceptionResNetV2, InceptionV3, Xception, and DenseNet with an emphasis on maize, cassava, and tomato diseases in tropical cultures. Furthermore, algorithms for plant detection like YoloV5, YoloV8, SSD, and FasterRCNN are evaluated. The findings highlight the dominance of Xception and DenseNet in classification, while YoloV5 performs exceptionally well in plant detection with a mean Average Precision (mAP) of 0.977 and an astounding 97% accuracy. This study emphasizes how cutting-edge methods have the power to transform agricultural disease diagnosis and reduce worldwide output losses