Desarrollo de un prototipo de sistema de detección de etiquetas impresas en sacos de balanceado mediante redes neuronales

Abstract

The detection and verification of labels on sacks is a significant challenge in maintaining the quality and safety of products in the current context of the balanced feed production industry. Historically, this process has been carried out manually, which is inefficient, costly, and susceptible to failures. The project emerges as an innovative solution to this scenario, automating and exponentially improving this process using neural networks and advanced artificial vision technologies. The project focuses on the implementation of a system based on the Jetson Nano development board and the YOLOv8 neural network model. These technologies enable real-time data processing and precise label detection, which is crucial in dynamic production environments. Incorporating a high-resolution camera is essential to capture detailed images, ensuring effective and reliable label detection. Beyond its specific application in detecting labels on balanced feed sacks, the project has a broad scope, with potential for adaptation to other industrial areas where precise product identification is key. The use of these advanced technologies not only improves operational efficiency but also opens doors to innovation in automation and quality control in various sectors.