Transfer Learning-Based Optimized Deep Learning Model to Characterize Mango Leaf Disease

Abstract

Human evolution has seen great advances in medicine, resulting in a complex way of life that is heavily dependent on natural resources. In this context, it is critical to protect the environment for plants, which provides society with priceless resources such as oxygen, nutrition, and health benefits. To solve this, emerging artificial intelligence approaches, especially deep learning, are being used to safeguard plant species from disease, such as the mango. Mango plants, often known as the "king of all species," have numerous benefits, including anti-cancer, antibacterial, and medicinal capabilities for respiratory and kidney conditions. Using a convolutional neural network (CNN) approach, the most commonly encountered diseases affecting mango plants are correctly detected. For agricultural workers, this proposed leaf disease detection method is quite useful because it automates the detection of leaf diseases in mango plants. 4000 photos of mango plant leaves divided into eight categories were used in the disease identification process. The outcomes of the comparison, analysis, and accuracy levels are fully presented, demonstrating the efficacy of the technique. This disease identification system, which makes use of CNN and deep learning techniques, not only helps to protect the health of mango plants but also advances understanding of the numerous diseases that affect them. The large dataset and accurate disease detection allow for a quick and focused response, minimizing the effects of various problems on the health of mango trees. Such improvements in automated plant disease diagnosis are important for preserving valued plant species like the mango and promoting sustainable agricultural practices. The anticipated model for mango leaf disease attained 99% accuracy.