Automated Sorting and Grading of Fresh Tomato Fruits Based on Image Classification

The sorting and grading of fresh tomatoes are critical in ensuring market value, reducing post-harvest losses, and enhancing quality standards. Traditional manual sorting methods are labour-intensive, inconsistent, and inefficient, necessitating an automated solution. This research involves the design and development of an automated system for sorting and grading tomatoes based on their ripeness levels: ripe or red tomatoes, unripe or green tomatoes, and partially ripe or yellowish-green tomatoes. The system uses image classification techniques, leveraging deep learning through a convolutional neural network (CNN) to classify tomatoes with high accuracy. A comprehensive dataset of tomato images will be collected and used to train and validate the model, which will then be deployed in a physical sorting system. Raspberry Pi will be integrated into the system for real-time image processing and control, providing enhanced computational power and connectivity compared to traditional microcontroller-based systems. The expected outcomes include a scalable, efficient, and highly accurate system capable of real-time classification and sorting, offering significant improvements over manual methods. Future recommendations include integration with harvesting machinery and expanded grading criteria for comprehensive agricultural automation.