Análise Comparativa de Algoritmos de Detecção de Objetos para Reconhecimento de Buracos em Vias e Estradas

Abstract

This work presents a comparative analysis using the mAP metricof the YOLOv7 and SSD algorithms applied to the field of computervision. The comparison is performed through the detection of potholeson highways and roads. In works defended by other authors,both techniques are considered for applications in computer vision.In this study, both models achieved promising results in potholedetection; however, it is observed that the YOLOv7 model reachedan mAP metric of 80%, while the SSD model obtained an mAP of73%. Thus, the results indicate that YOLOv7 is more efficient andaccurate in pothole detection in this specific context.