Automated Sorter and Grading of Tomatoes using Image Analysis and Deep Learning Techniques

Many sorting machines were made in the past that classifies the ripeness of tomatoes. These machines do not classify the size and quality of tomatoes like physical damage, defects, and bruises of plants, resulting in the traditional checking of the quality and size estimation. Based on the Philippines Statistics Office, the Philippines produces many tomatoes that become the driving force to push through this study. This study aims to innovate the existing method of sorting and grading tomatoes depending on the size and quality, including the color, defects, physical damage, and bruises. Specifically, this study aims to create software and hardware that can analyze tomatoes' grades based on color and classify the fruit according to size. Also, to conduct a statistical analysis of the efficiency of the system. The process starts from capturing an image of a tomato, which will be analyzed by applying the image processing techniques such as deep learning algorithms to determine the quality and the Canny-Edge Detection to categorize the size in small, medium, large, and jumbo. The Conveyor is controlled by an Arduino microcontroller transporting the tomatoes into the spot where the tomatoes' image is captured. This study results in an overall efficiency of 88% after testing multiple batches of tomatoes of different sizes and quality.