Deep Learning CNN Models for Diseases Classification in Cauliflower Leaves

Cauliflower, a widely consumed vegetable valued for its nutrition in cooking, encounters significant agricultural difficulties because of the presence of various diseases that have an adverse impact on its quality and production. Early detection of these diseases is essential for timely plant treatment and increased production. This study presents a novel approach for detecting cauliflower leaf disease using deep learning techniques, taking use of the advances in deep learning for image classification. The study utilizes a dataset consisting of images of healthy leaves and affected leaves by widespread diseases such as Alternaria Leaf Spot, Black Rot, Cabbage Aphid, and Cabbage Looper. A pre-trained convolutional neural network (CNN) architecture is optimized and customized for this particular study in order to achieve disease classification. The proposed study has concentrated on fine-tuning the hyperparameters for the commonly used models such as NASNet Mobile, ResNet50, and Inception V3. The dataset used in this research contains 729 images of cauliflower leaves collected manually with the aid of a mobile phone camera from different farm fields in the Bhaktapur district of Nepal. Several performance metrics, such as accuracy, precision, and recall were used to evaluate the model’s performance. The experimental result shows that the ResNet50 has better performance with an accuracy of 93.47% compared to other models NASNet Mobile and Inception V3.