Real-time FFB ripeness detection using IoT-enabled YOLOv8n on Raspberry Pi 4 edge devices for precision agriculture

This paper presents the development of an edge device for cost-effective implementation in agricultural environments. Experimental evaluations demonstrate accuracy and real-time performance, showcasing its potential for adoption in the industry. The proposed system provides a reliable tool for timely and accurate monitoring of fresh fruit bunch (FFB) ripeness, facilitating optimized crop management practices. The system employs the YOLOv8n model, renowned for its efficiency in real-time object detection tasks, and is adapted to run on the resource-constrained Raspberry Pi 4. To ensure seamless operation on edge devices, model optimization techniques such as quantization and hardware acceleration are implemented, enabling rapid decision-making based on live data feeds. A dataset comprising 4,194 annotated FFB images was utilized, with a [3,681:348:165] training-validation-test split. Performance evaluation demonstrated an average precision of 0.898 and a mean average precision (mAP) of 0.952. The system potentially enhances yield quality and sustainability while supporting data-driven decision-making in precision agriculture.