Sistema de teledetección aérea para control de agentes patógenos y enfermedades en cultivos de brócoli con el uso de visión artificial

Abstract

Broccoli is one of the agricultural products in Ecuador with the highest rate of imports worldwide, due to this the development of the crop must be routinely inspected to counteract the curses that harm its production such as pests or diseases and preserve its high rate of productive quality. For this reason, the present research project is oriented to carry out inspection of broccoli crops by air with pre programmed flight plans that help to analyze the health status of the crop to be treated on time. The design and implementation of the aerial remote sensing system for the control of pathogens and diseases in broccoli crops is developed through the use of the YOLO v5x artificial vision algorithm for deep learning of the system in various production circumstances. The aerial remote sensing system uses a drone that helps to move the active and passive sensors that interact in the system. This drone is autonomous thanks to the implementation of a GPS module on its flight control board, which helps planning grid-type pre-programmed flights. This system receives the analysis of the health status of the broccoli crop by capturing images through a high-definition camera incorporated into the drone. Each image is captured in a time frame of 10 seconds to carry out its respective processing in real time through a graphical interface programmed in Python. The image consists of positioning data such as altitude and latitude obtained through a geolocation system that was implemented in the drone. These data help to locate the specific point where the images were captured and interpret the results obtained in the analysis of the culture for each one of them. The results of the image processing are stored in a database, for the autonomous training of the artificial vision algorithm and improvement of the prototype in the detection of false positives and false negatives