Plant Disease Diagnosis: A Deep Learning Approach

Plant diseases have a considerable impact on agricultural crop output and quality. Automated disease identification using computer vision algorithms can help in early diagnosis and containment of disease transmission. In this study, we looked into the use of deep learning models for the detection of plant diseases, specifically the Xception net for feature extraction and the Vision Transformer for classification. With over 50,000 photos of 38 different plant species representing 14 different disease classifications, the Plant Village dataset was used. Our goal was to outperform earlier research on this dataset in terms of results. In order to detect plant diseases using deep learning models, we reviewed the literature on recent studies and research in this area. We chose Xception net and Vision Transformer as our models for feature extraction and classification, respectively, based on our investigation. The Plant Village dataset was gathered and pre-processed, and training and validation sets were created from it. After that, we used the dataset to train and assess the models, using the evaluation criteria of accuracy, precision, and loss. According to our experimental findings, the combination of Xception net and Vision Transformer outperformed earlier studies on this dataset in terms of accuracy on the validation set. We examined the variables that affected the models’ performance and talked about the implications of our study for agricultural and food security. Our study contributes to the development of automated disease detection tools that can aid in early illness detection and disease prevention, ultimately enhancing crop yields and quality.