USING CONVOLUTIONAL NEURAL NETWORK TO DIAGNOSE DISEASES ON MANGO LEAVES

The objective of this research is to apply Convolutional Neural Networks (CNNs) for the accurate identification of seven common diseases on mango leaves: Anthracnose, Bacterial Canker, Cutting Weevil, Die Back, Gall Midge, Powdery Mildew, and Sooty Mould. A total dataset of 4,800 images was collected, with 600 images for each disease type. The entire image dataset was divided into three sets: 3,840 images for the training set and validation set; 960 images for the test set. A deep learning model, utilizing a CNN architecture, was developed and trained to classify mango leaf images. The model achieved a significant accuracy of over 97.60% on the test set, demonstrating its strong capability to recognize and differentiate between various mango leaf diseases. This high model accuracy enables early and precise diagnosis of mango leaf diseases, facilitating timely intervention measures and minimizing crop losses. This can lead to improved agricultural management practices, enhanced crop yields, and reduced reliance on chemical treatments. Furthermore, developing an automated disease detection system can assist farmers and agricultural professionals in making informed decisions. Keywords: Mango leaf diseases; deep learning; convolutional neural network (CNN); image classification; plant disease detection