Optimization of Mango Leaf Disease Detection Using a Hybrid Machine Learning Approach

Mango cultivation is essential in northern Côte d'Ivoire, substantially contributing to the national economy. However, microbial diseases affecting mango leaves represent a significant challenge for farmers. Early detection of these diseases is crucial for effective crop management and plantation protection. Infrastructure constraints limit the ability to detect mango leaf diseases early. To deal with this challenge, a new approach based on artificial intelligence has been developed to detect and classify mango leaf diseases. The study is based on two methodological approaches. Firstly, we use machine learning algorithms, including Random Forests, Support Vector Machines (SVMs), and eXtreme Gradient Boosting (XGBoosting). In parallel, a second approach incorporates Convolutional Neural Networks (CNNs) to extract complex visual features from leaf images. These features are combined with the three machine learning algorithms mentioned above for classification. The results show that the second approach, which combines CNNs with machine learning algorithms, outperforms the first. In particular, the accuracy of the second approach, with CNNs combined with SVMs, stands out, achieving the highest accuracy scores. The performance of the VGG16-SVM, ResNet-SVM, and VGG19-SVM models are evaluated with high precision and accuracy scores, respectively. These results offer promising prospects for the practical application of these techniques in fields such as medical image classification, materials analysis, and other areas where spectroscopy is used as a classification tool.