Aplicación de técnicas de visión artificial para la contabilización de flores en cultivos agrícola

Abstract

Abstract: This thesis project aims to develop an artificial intelligence (AI) application for automated flower detection and counting. Its application is intended for agricultural crops, using computer vision techniques to optimize a process that is traditionally performed manually. The goal of counting is to reduce the time spent when performing the activity manually. As a solution, computer vision was implemented with the YOLOv5 model, which allows us to custom-train its detection. The finished application will allow us to upload videos containing flower captures in the fields, so that the video identifying and counting the flowers will be returned as a response. The results obtained demonstrated that the application is efficient and precise, achieving better results than manual methods, achieving greater accuracy even in low-light conditions. Therefore, we can affirm that this work provides us with an innovative solution that contributes to the development of agriculture that looks to the future with artificial intelligence.