Utilizing TFLite and Machine Learning for the Early Detection of Mango Leaf Disease: An Automated Flutter Application

Abstract

The mango, one of the most important tropical fruits in the world economically, is constantly in danger of going extinct due to a number of diseases. In order to achieve optimal production and quality, prompt intervention and efficient management depend on the early and accurate detection of these diseases. Expert visual inspection is a common component of traditional illness diagnosis techniques, although it can be laborious, subjective, and prone to human mistake. In this research work, we applied Machine Learning techniques to create an automated system for detecting mango disease. The application uses a TFLite model, which was trained on a huge set of images of mango leaves with the following classifications: Powdery Mildew, Bacterial Canker, Black Sooty Mold, Cutting Weevil, Die Back, Anthracnose, and Healthy Leaves. The mango leaf disease dataset is obtained from Kaggle with 300 images. We have evaluated the model with 10 epochs and achieved 100% of accuracy.