Cauliflower Disease Detection Using a Deep Learning Approach

Abstract

Cauliflower, a significant crop in global agriculture, is susceptible to various diseases that can severely impact yield and quality. Accurate detection and classification of the diseases are crucial for effective crop management and sustainable agricultural practices. This research paper explores the application of deep learning techniques for the detection and classification of cauliflower diseases from images. The study investigates several state-of-the-art deep learning architectures, including CNN, EfficientNetB3, EfficientNetB7, EfficientNetV3, MobileNetV3, ResNet32, and ResNet50. These models were trained on a custom-collected dataset comprising five classes: Cauliflower Healthy, Cauliflower Healthy Leaf, Cauliflower Leaf Black Rot, Cauliflower Leaf Red Spot, and Cauliflower Spot Rot. Comprehensive image preprocessing, including augmentation and contrast enhancement, was employed to improve model robustness and generalization. The performance of each model was evaluated using metrics such as precision, recall, and accuracy. MobileNetV3 achieved the highest accuracy of 99%, demonstrating its superior ability to distinguish between healthy and diseased conditions. This high accuracy underscores the potential of deep learning in supporting precise agricultural interventions.