Desarrollo de un sistema de reconocimiento de imágenes para la clasificación de limones usando Raspberry PI 4 con Python

Abstract

The objective of this project is to build an automated conveyor belt using artificial vision to classify lemons into green, ripe and rejected categories, resulting in a precise and efficient classification. The conveyor belt is equipped with sensors and cameras that capture images of the lemons in motion, which are processed by an automatic learning model previously trained with Google's Teachable Machine digital tool. This automatic model classifies lemons by their color. The project process begins with the capture of real-time images as the lemons move along the conveyor belt. These images captured by the camera are sent to an identification process that uses the artificial intelligence model previously trained with images of lemons' conditions. An analysis of the mechanical and electronic components is carried out to ensure a reliable and efficient process. In additional, different image classification methods and techniques are investigated to ensure that the model is able to handle variations in conditions such as brightness, as this can affect the identification of the model used. As a result, there is an automated conveyor belt that effectively sorts lemons by using digital tools and electronic components that are easily available on the market. This project innovates the way of sorting lemons in the agricultural industry that requires automated solutions to achieve better performance in the management of their operations.