Disease detection from potato leaf images using deep learning methods

This thesis aims to detect diseases from potato images using deep learning methods. In the study, a large and comprehensive image dataset of healthy and various potato diseases was used. Models were developed to detect potato diseases using different Convolutional Neural Network (CNN) architectures and hybrid models. The developed models were trained using different parameters and datasets and evaluated using metrics such as accuracy and precision. Common diseases seen in potato plants (late blight, early blight) were detected and the performance of the models was increased using image preprocessing techniques. This study aims to show that deep learning methods can be used effectively in the detection of potato diseases and to contribute to previous studies in this field. In the study, libraries, GPUs, processors, natural language processing models and Google Colab platform, which are indispensable components of experimental environments, were examined and the PillantVillage dataset was used. Images were tested with four different ResNet models and evaluated with various performance metrics. It is thought that the findings obtained can provide important information for disease management and productivity increase in potato cultivation. Disease detection from images with artificial intelligence can lead to innovations in the field of agriculture and can also contribute to machine-human interaction. Our study emphasizes the success and importance of deep learning models in the field of image extraction, ResNet deep learning models.