Analysis of CNN Models with Transfer Learning in Mango Diseases Detection

Mangoes are a valuable fruit and a source of antioxidants because they contain phytochemicals like carotenoids and polyphenols. But mangoes are susceptible for various diseases such as bacterial canker, anthracnose, and powdery mildew. In this study, the main approach for detecting diseases in mango crops—with a focus on diseases that have a major impact on production and quality—is Convolutional neural networks (CNNs). We preprocesses mango images by encoding class labels and standardizing pixel values using a Kaggle dataset in order to maximize model performance. Through transfer learning, several CNN architectures—including VGG16, VGG19, ResNet50, DenseNet121, and AlexNet—are compared, and high accuracy rates are attained. The models ResNet50 and DenseNet121 have shown their efficacy in identifying mango leaf diseases by achieving the maximum accuracy, up to 100%. This study highlights how CNN models may be used to enhance early disease identification in mango crops, perhaps leading to increased agricultural output and better financial results for farmers.