Identification of Musaceae Species using YOLO Algorithm

A significant amount of research on agriculture has made use of image processing. The vast majority of the currently conducted research is only focused on the identification of fruit and the detection of diseases that affect fruits. There are very few, if any, studies that concentrate on the process of determining the species of the fruit itself. With this, the researchers came up with the idea for a method that can distinguish between the four different species of banana, which are known as Lakatan, Latundan, Cardaba, and Senorita. The system that was built makes use of an algorithm known as You Only Look Once (YOLO), which can detect and identify multiple classes in a short period. Raspberry Pi 4B 8GB and Raspberry Pi Camera Module v.1.2 are the two primary components that make up the hardware of the system. The accuracy of the system was improved by applying several image pre-processing methods to the images, such as the conversion of RGB to grayscale, image binarization, morphological transformations, changing of background color, and image resizing. To test the effectiveness of the system, the researchers used a confusion matrix and performed calculations to determine both the system's accuracy and its rate of incorrect categorization. The system's overall accuracy is 93.65%, which means that 241 out of 252 photos were accurately classified as their own species.