Tomato Ripeness and Size Classification Using Image Processing
Jericho Legaspi, John Raphael Pangilinan, Noel B. Linsangan
Abstract
The tomato (Lycopersicum esculentum Miller), locally known as “Kamatis” in Filipino, is one of the fruit vegetables grown locally in the Philippines. The sortation and classification of both its size and ripeness are primarily based on the manual sortation of people, which poses a problem regarding accuracy and consistency. This study developed a system that classifies tomatoes' ripeness level and size using image processing. InceptionV3 was used in the ripeness classification, while OpenCV was utilized in the size classification. The system is composed of the Raspberry Pi, Raspberry Camera v1.3, and an Ultrasonic sensor for its hardware components. Seventy (70) tomato samples were utilized (ten per ripeness stage and ten as empty) to test the ripeness under seven classifications, and fifty (50) were used (ten per size classification and ten as empty) to test size under five classifications. A confusion matrix table was applied to analyze the result garnered from the testing stage. The ripeness classification system garnered an accuracy score of 92.86%, and the size classification has an accuracy of 96%.