Identifying Mango Leaf Diseases with Advanced Deep Learning Approaches and Convolutional Neural Networks

Examining whether any leaf is diseased or not by visual inspection of the naked human eye is often unreliable and incorrect. Technological developments in the form of techniques like deep learning have proved to be extremely useful in agricultural threat detection by the detection of leaf diseases. It is a well-known fact that whenever a plant is affected by any disease, its first symptoms can be identified in the leaves of the tree. Research done on mango tree leaf infections using the EfficientN et Model and FastAI framework has been highlighted in the following paper. The FastAI framework provides a better ensemble for mango leaf disease recognition(MLDR). Images of new leaves entered by the user after proper cropping and focusing can be entered as input to know if the plant is diseased or healthy. The average error rate of the model's classification and detection of illnesses is 0.4%. To achieve such drastic results, a dataset of 12046 images was used that demonstrated nine different classes(1 healthy and 8 different types) of infected leaves. The proposed model having trained by the same dataset demonstrated superior performance and proved to be most proficient and efficient. The current pre-trained EfficientN et model outperformed the current state-of-the-art techniques as well as other test models by achieving an F1 Score of 0.9960., Accuracy of 98.97% and a Precision Score of 0.9905. Implementing this strategy can enhance mango production., satisfy international demands and significantly improve the management of diseases affecting mango leaves.