Intelligent Fruit Inspection System: Developing a YOLO-based Model for Identifying Defects on Plums Surface
Keywords: Agriculture, Artificial intelligence, Computer vision, Deep learning, African pears, YOLOv5, YOLOv8
TL;DR: AI solution for African pear quality assessment using YOLOv5 and YOLOv8 with mean average precision of 88.2% and 89.9% respectively; YOLOv8 deployed in web app, first intelligent system for African plum inspection.
Abstract: Agriculture is the backbone of Africa’s economy, with over 60% of the labor force of over 1.2 billion people largely depending on it for their livelihoods. However, the gap between agriculture and technology in Africa continues to widen, creating a need for innovative solutions to improve productivity and access to high-quality agricultural products. One such problem is manually assessing the quality of
agricultural products, which is especially time-consuming and tedious when done on a large scale. To address this challenge, we developed an artificial intelligence solution using YOLOv5 and YOLOv8 algorithms to assess the quality of African pears. We collected, from three regions in Cameroon, a dataset of 2892 damaged and good African pear surfaces. Our YOLOv5 and YOLOv8 models achieved
mean average precision scores of 88.2% and 89.9% respectively. The proposed YOLOv8 solution has been deployed and runs on a web application. To the best of our knowledge, this is the first intelligent system for inspecting African plum quality.
Submission Category: Machine learning algorithms
Submission Number: 45
Loading