Reducing Post-Harvest Losses in Sri Lankan Pineapple Farming Through YOLOv8-Based Ripeness Detection

Post-harvest losses remain a critical challenge for smallholder pineapple farmers in Sri Lanka, primarily due to inaccurate ripeness detection methods. Traditional manual assessments are prone to human error, leading to significant economic losses and reduced fruit quality. This research presents a real-time YOLOv8-based pineapple ripeness detection system to enhance harvesting accuracy and minimize post-harvest losses. A dataset of pineapple images was collected and preprocessed, and the YOLOv8 model was trained to classify pineapples into four ripeness stages: unripe, partially ripe, ripe, and overripe. The system was integrated into a web-based application, allowing farmers to upload images or capture them via webcam for immediate ripeness evaluation. The model achieved a precision of 92%, recall of 89%, and an F1- score of 90.5%, demonstrating its reliability in real-world conditions. Performance tests confirmed the system’s efficiency, with an average detection time of less than 100ms per image. The proposed solution empowers smallholder farmers by providing an accessible, cost-effective, and scalable tool to optimize harvesting decisions, reduce waste, and enhance profitability. Future improvements include drone-based crop monitoring and a mobile application for enhanced usability and scalability.