A Deep Learning Approach for Cauliflower Leaf Disease Detection Using Image Analysis

Abstract

Cauliflower is a major vegetable crop all over the world, but in several case leaf diseases like black rot, downy mildew and bacterial leaf spot have a substantial impact on the output and quality. For efficient crop management and the reduction of financial losses, there needs of early and precise detection of these diseases. Conventional disease detection techniques are depend on professional manual inspection. This thing is labor-intensive, time-consuming and prone to human mistake. The promising approaches to automated plant disease diagnosis are provided by recent developments in the world of deep learning and computer vision. Over here deep learning-based method for image analysis-based cauliflower leaf disease identification. The given approach accurately classifies cauliflower leaf illnesses by utilizing convolutional neural networks and VGG16 , this are cutting-edge deep learning technique. To improve the model performance, a dataset of pictures of cauliflower leaves healthy and diseases both was gathered and preprocessed. To improve dataset and avoid overfitting, data augmentation methods like rotation, flipping, horizontal, blur and noise add were used. Transfer learning was used to train, test and validation the CNN model utilizing pre-trained architectures like ResNet50, ResNet152, VGG16, DenseNet169 and CNN that were optimized for the categorization of cauliflower illness. On the basis of experimental results, the suggested deep learning model is able to do diagnose various diseases of cauliflower leaves with good accuracy, precision, recall and F1-score. As the advantage of CNN-based techniques in managing image features and enhancing detection reliability is demonstrated by a comparison with conventional machine learning techniques. By apply and contrast many deep learning models, such as CNN, ResNet152, DenseNet169 and VGG16 for the identification of cauliflower disease. VGG16 performed as 96% and CNN also has 75%. On other hand, all other models doesn’t have the good one. The findings show that VGG16 is a dependable option for agricultural applications due to its exceptional capacity for precisely diagnosing plant diseases. Accordingly, deep learning based image can be analyzed greatly improve the effectiveness and precision of detecting cauliflower illness, providing a scalable and affordable precision agricultural solution. Future research could include extending the dataset to include more disease types and climatic variables, as well as combining the model with mobile applications for field deployment.