The Darknet Framework YOLOV7 is Used for Pineapple Fruit Ripeness Classification

In this study, we developed and implemented an object detection method to distinguish between ripe and unripe pineapples. We used a pre-trained object detection model using YOLOv7 and BlendMask techniques and finetuned it on the MS COCO instance segmentation dataset for 30 epochs. Through the preprocessing process, we resized the images to the specified target size and normalized the pixel intensities. We also created bounding boxes to mark objects in the images based on computed coordinates using object scores and attributes. During the prediction stage on two sample images, we accurately predicted that one image contained a ripe pineapple with a confidence level of 56%, while the other image contained an unripe pineapple with a confidence level of 77%. These results demonstrate that our developed object detection method is effective in differentiating the ripeness of pineapples. This research contributes to the development of fruit classification applications based on ripeness levels. The implemented object detection method can serve as a foundation for the development of more advanced systems in automatically identifying and segregating ripe and unripe fruits. As a result, this study has the potential to enhance efficiency and accuracy in the agricultural and fruit processing industries.