Application of a Neural Network to Feed Bag Tag Detection Systems

This paper describes the implementation of an advanced artificial intelligence-based system for the detection and labeling of feed sacks, using the Jetson Nano development board and the YOLOv8 neural network model. This system employs computer vision techniques to accurately identify the sacks and their side labels. Ensuring the presence of a label on each sack is crucial for product monitoring and quality control of the labeling process. The methodology involves a rigorous model training process, using data captured in real production conditions, and system configuration to optimize detection performance. The results obtained demonstrate the system’s robustness and efficiency, with an accuracy rate of 98.84% in controlled conditions and 96.42% in a real production environment. This underscores the system’s practicality and its potential to significantly enhance logistics and quality control processes in the shrimp feed industry. The technology’s implementation can lead to improved product traceability and operational efficiency throughout the supply chain, potentially reducing costs, minimizing errors, and increasing end-customer satisfaction.