A Novel Approach for Tomato Quality Evaluation: Computer Vision in Maturity Assessment and Disease Detection

Tomatoes (Solanum lycopersicum) are one of the most widely consumed fruits globally. Fruit maturity and disease estimation is a criticalfactors in determining its quality and marketability. Traditional maturity assessment requires huge labor, slow, subjective, and time-consuming activities. To address these challenges, the agriculture industry is subjected to automation using computer vision models. Accurate prediction of tomato maturity and diseases is important to decide on harvesting time. Several features such as color, shape, texture, and size are considered to predict the different maturity stages of fruits using deep learning and machine learning models. Assessing the maturity level and disease detection in the early stage of tomatoes is crucial. The proposed work explores the application of the VGG16 architecturefor tomato maturity classification and disease detection. The study leverages transfer learning with VGG16focusing on feature extraction and model optimization to achieve accurate classification. The datasets used in this work are self-created using the images that are collected from agricultural farms. The dataset consists of image samples of six different maturity stagesand diseased fruit images. The experiment results in an accuracy range between 95%-99% for six maturity stages of tomatoes and disease prediction resulted in 90%-97% accuracy.