Shell Texture-based Identification of Damaged Tamarindus Indica Using YOLOv5

The objective of this research was to create a computer vision system utilizing the YOLOv5 algorithm and identify damaged Tamarindus Indica by analyzing the external characteristics of their shells. Despite the widespread use of tamarind fruit in culinary preparations, research on identifying healthy tamarinds remains limited. Using a capturing chamber equipped with a USB camera and Raspberry Pi 4B, we captured 1050 images for training and testing the model. The results demonstrated that the developed YOLOv5 model showed an accuracy of 93% in detecting fungal damage, 91 % in detecting cracked damage, and 96% in detecting healthy tamarinds. Overall, the accuracy rate of the model was 93 % in detecting damaged and healthy tamarinds, which shows the great potential of the proposed method for use in the quality control of tamarind products.