Author: Lilou Gras
Date: 04/06/2023
email: [email protected]
University of Tartu
Plant diseases can have detrimental effects on crop yield and quality, leading to significant economic losses. Moreover, the misdiagnosis of diseases can result in the usage of the wrong and potentially dangerous pesticide. Using machine learning techniques, specifically deep learning algorithms, can help farmers to diagnose their plant with a simple picture. In this study, pictures of healthy and infected apple tree leafs were provided with their corresponding label. With the help of DenseNet121, a popular convolutional neural network architecture, alongside additionnal techniques I attempted to classify the different diseases of the apple tree. Experimental results demonstrated that the proposed approach achieved great classification accuracy, and showed the potential of deep learning in disease identification. However, this work could still use some improvements, even if it is more reliable than the traditional method of farmers trying to diagnose the pest by themselves. Future work may involve augmenting the dataset, exploring other deep learning architectures (ResNet, Yolo, EfficientNet...), and deploying the model in real-world scenarios to further validate its effectiveness and scalability.