Intelligent Detection of Apricot Ripeness using YOLOv8: Embedded Application on Raspberry Pi 5 for Precision Agriculture

Abstract

This thesis presents an embedded system for intelligent apricot ripeness detection using YOLOv8 on Raspberry Pi 5, aiming to modernize fruit quality assessment in Algerian agriculture. By creating and preprocessing a custom dataset, training and evaluating several YOLOv8 models, and optimizing for real-time edge deployment, the system enables accurate, fast, and objective classification of ripe and unripe apricots. The solution addresses challenges of labor scarcity and post-harvest losses, offering farmers a practical tool for better harvest timing and sustainable orchard management. This work demonstrates the potential of deep learning and embedded AI to bring tangible benefits to precision agriculture.