| Summary: | Durian, a renowned fruit in Southeast Asia, particularly in countries like Malaysia, Thailand, and Indonesia, poses a challenge during its season as the manual classification of a large stock of durians based on grade and quality becomes a laborious task for sellers. This project aims to streamline the process by employing image processing techniques for the classification of durians into defect and non-defect categories. The methodology involves image collection and image filtering analysis using Gaussian and Median filters, followed by applying Canny edge detection techniques to identify the durian region. Subsequently, classification algorithms based on pixel connection are deployed to distinguish between defect and non-defect durians. The obtained results reveal comparable accuracy and precision rates for both defect and non-defect durian images, standing at 87% and 75%, respectively. This project successfully demonstrates the feasibility of automating durian classification through image processing methods.
|
|---|