Algoritmo de diagnóstico para evaluar parámetros de calibración en los sensores IoT Xovis con el propósito de sistematizar el proceso de calibración en la empresa Followup, basado en procesamiento de imágenes

Abstract

FollowUp Company provides a people counting service using Xovis brand IoT sensors. To ensure the accuracy of the service, a periodic calibration process of the sensors is carried out. This process is based on analyzing the margin of error between people entering and leaving the monitored area. If this value exceeds 5%, the Xovis API is accessed to analyze the sensor's real-time image. The analysis utilizes height maps, stereo image maps, start-stop maps, and an old image of the sensor to determine whether calibration is necessary, if there is another issue at the sensor's location, or if another intervention is required. The Python library OpenCV is used for image processing. The algorithm is hosted in a script that takes the MAC addresses of the sensors as input and outputs a new Excel list of sensors that require calibration.