From Pixels to Prognosis: Exploring Convolutional Neural Networks for Cassava Leaf Disease Diagnosis

Cassava is a major crop in many areas, but it is susceptible to a number of illnesses that can have a terrible impact on crop productivity and food security. For successful disease treatment and prevention, these diseases must be accurately and quickly detected. In this paper, we suggest a deep learning-based approach for diagnosing cassava leaf disease. We create a modified convolutional neural network architecture for feature extraction and classification that includes parallel pathways, multiple convolutions, max pooling, and global average pooling. Cassava leaf images from a large dataset, including both healthy leaves and several disease kinds, are used to train the model. Through extensive experimentation and evaluation, the proposed model demonstrates high accuracy and robust performance in classifying cassava leaf diseases. The proposed model has also achieved a training accuracy of 98.54%, validation accuracy of 97.84% and a testing accuracy of 98.08%. The suggested approach has enormous potential to aid farmers and researchers in early disease diagnosis, enabling prompt treatments and efficient crop management. Utilising cutting-edge technology to solve current problems and enhance sustainable agricultural practices is important, as demonstrated by the application of deep learning techniques in agriculture.