Abstract

BOGRAN ORTIZ, Lester Yonabel  and  MARTINEZ HERNANDEZ, Jairo Jonathan. Comparison of detection models for objects and people in closed spaces with public access. Universidad y Sociedad [online]. 2023, vol.15, n.4, pp. 661-672.  Epub Aug 12, 2023. ISSN 2218-3620.

This work aims to present a comparison of different object detection techniques and determine which of them is the most suitable to detect people in real time and thus efficiently control the capacity of people in closed public spaces to help prevent the spread of COVID-19. In this study, solutions based on neural networks such as R-CNN, YOLO and SSD, as well as non-neural solutions such as SVM and HOG, have been tested. We worked with models trained with the MS COCO dataset, and the Wisenet dataset was used to evaluate the different models. All models were tested in three different aspects, these being precession, speed of inference and error. In tests, the YOLO model demonstrated higher performance than the others, while the SSD implementation demonstrated the lowest performance in most tests.

Keywords : Object detection; Computer visión; Convolutional neural networks; HOG.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )