Implementación de un modelo de inteligencia artificial para la clasificación de materiales metálicos y no metálicos en una banda transportadora

Abstract

The project is based on a sorting system for metal and plastic waste based on artificial vision, implemented on a conveyor belt. An image processing approach is used that involves capturing, preprocessing and segmenting visual data to train a deep learning model. The system is designed to operate in real time, allowing the identification and categorization of materials, whether metallic or plastic. For this purpose, an Nvidia Jetson Xavier NX is used, which performs the inference of the optimized model, ensuring high efficiency in object recognition. The project also incorporates an electromechanical actuation system, consisting of a set of dual-channel relays and a controller that allows automatic sorting by activating separation mechanisms. The integration of Arduino Nano ESP32 facilitates communication between the different modules, ensuring a coordinated response in the execution of the process. Finally, validation tests are carried out to analyze the performance of the system in terms of accuracy and response times, obtaining results that demonstrate its viability for industrial applications in the field of automated recycling.