Plant Disease Classification Through Image Representations with Embeddings

In response to the significant challenges faced by the agricultural sector due to climate change, this research focuses on enhancing plant disease detection and management through advanced computer vision techniques. The study introduces an innovative Artificial Intelligence (AI) model leveraging image classifications and embeddings to predict plant diseases, utilizing a public dataset comprising 27 combinations of plants and diseases. This research integrates Natural Language Processing (NLP) techniques to generate embeddings from images, which are then utilized to train a classification model based on classical machine learning algorithms. The proposed system employs a novel approach by transforming a typical computer vision problem into a feature-based classification problem, allowing the use of diverse machine learning models. The efficiency of the model is demonstrated through its ability to accurately classify various plant diseases, with the best-performing model achieving promising results in terms of accuracy and reliability. The research highlights the potential of combining technological innovations with agricultural expertise to address the complexities of modern agriculture. Future directions include refining the model by exploring different embedding techniques and increasing the dataset using data augmentation methods, to further improve the system’s performance.