A New Deep and Ensemble Learning-Enhanced Approach for Diagnosing Mango Leaf Diseases

Mango cultivation plays a vital role in generating billions of dollars in the global market, particularly from leading producers such as India, China, and Thailand. However, a major issue that has been affecting mango trees globally is mango leaf disease which resulted in an overall decrease of yield by 30%. It is essential to identify the disease at its initial stage in order to mitigate the risk. Various contemporary techniques are employed for disease identification, deep learning performs exceptionally well among them. Numerous researchers have proposed and developed deep learning-based algorithms aimed at categorizing these diseases. This study aims to introduce a novel ensemble-based deep learning model designed for multiclass classification to identify various types of leaf diseases. The model training and evaluation was done on a dataset comprising 19,570 images across eight categories, attaining an impressive accuracy of 99.12%, along with recall, F1-score and precision values of 1, 0.99, and 0.99, respectively, underscoring the model's robustness. This model is highly efficient and readily deployable as a practical application, offering valuable assistance to both farmers and large-scale agricultural firms.