Mestrado em Engenharia Elétrica
URI Permanente para esta coleção
Nível: Mestrado Acadêmico
Ano de início:
Conceito atual na CAPES:
Ato normativo:
Periodicidade de seleção:
Área(s) de concentração:
Url do curso:
Navegar
Navegando Mestrado em Engenharia Elétrica por Autor "Almonfrey, Douglas"
Agora exibindo 1 - 2 de 2
Resultados por página
Opções de Ordenação
- ItemDetecção de Armas de Fogo em Imagens Baseada em Redes Neurais Convolucionais(Universidade Federal do Espírito Santo, 2021-04-30) Cardoso, Guilherme Vinícius Simões; Ciarelli, Patrick Marques; https://orcid.org/0000000331774028; http://lattes.cnpq.br/1267950518719423; https://orcid.org/0000-0003-0898-9186; http://lattes.cnpq.br/7411793218517484; Samatelo, Jorge Leonid Aching; https://orcid.org/0000-0001-7679-4132; http://lattes.cnpq.br/5049258096050209; Almonfrey, Douglas; https://orcid.org/0000-0002-0547-3494; http://lattes.cnpq.br/1291322166628469; Vassallo, Raquel Frizera; https://orcid.org/0000000247623219; http://lattes.cnpq.br/9572903915280374The demand for weapons has grown along with crime rates, being a contemporary problem that haunts several countries. In Brazil, possible changes are being discussed to make the ownership and possession of firearms more flexible, dividing opinions and generating a huge discussion on the subject. This has motivated scientists to devise solutions that can assist in public security in general. In an attempt to find ways to minimize this problem, a research was carried out on the main work related to the classification and detection of firearms, aiming to obtain information on the main techniques used. Thus, in this work is proposed a methodology for the detection of firearms in images using convolutional neural networks. Recent work has used object detectors based on these networks and presented relevant results. Therefore, this work proposes a methodology for detecting weapons using an object detector, called YOLO (You Only Look Once), and an architecture based on convolutional neural networks. Two approaches were taken to evaluate the proposed methodology, taking into account three threshold values for IoU. The first approach, compared with the results found in the literature, points to an improvement in the results, where an accuracy of 93, 67% and a F1 of 93, 05% was achieved, which represents a growth greater than 10% in accuracy and a slight improvement of almost 2% in the F1 metric. The second approach follows the same methodology, but applies a different initial step, where the object detector is modified and used to mark a database and compose a new labeled one. Such approach had a positive impact on the results, where there was an increase in accuracy and almost 4% in the F1 metric. In the three IoU values evaluated, the best one has an accuracy of 89, 91% and, in the same configuration, points to a F1 of 94, 54% with a confidence of 58%. These results show that the proposed methodology is promising to be applied for firearms detection in images.
- ItemUso da constância de cor na robótica móvel(Universidade Federal do Espírito Santo, 2011-07-21) Almonfrey, Douglas; Schneebeli, Hans-Jörg Andreas; Vassallo, Raquel Frizera; Salles, Evandro Ottoni Teatini; Stemmer, Marcelo RicardoThe color captured by a camera is function of the scene illumination, the reflective characteristics of the surfaces in the scene, the photosensors in the vision systems and mainly the processing made by the brain. Due to this processing performed by the brain, humans show the color constancy phenomenon: the color of a surface is perceived as the same regardless of the environment illumination conditions. However, the variation in the scene illumination implies a change in the color value of a surface registered by an artificial vision system. In the literature, defining surface descriptors that are independent of the illumination is known as color constancy problem. One solution to this problem is to obtain the reflective characteristics of the surfaces apart from the information of the scene illumination. Another approach to solve the color constancy problem is to convert the colors of the surfaces in the image so that the surfaces appear to be always under influence of the same standard illumination. Independently of the chosen approach, this is a hard problem to solve and most existing theories are applied only to synthesized images while others present a limited performance when applied to real images of environments under uncontrolled illumination. Due to the absence of the color constancy phenomenon in artificial vision systems, many automatic systems avoid the use of color information obtained from images captured by these systems. Besides that, the solution of the color constancy problem is also desired by the consumer photography industry. In this context, this work addresses the solution of the color constancy problem using an algorithm based on the color correction method presented in (KONZEN; SCHNEEBELI, 2007a). This algorithm corrects colors of a scene captured under unknown illumination so that the scene appears to have been captured under the influence of a standard illumination. If the scene illumination is always the same, the colors of the images show color constancy. This conversion between illuminations is performed by knowing the colors of some points in the scene under the influence of the standard illumination. Finally, we analyze the color constancy algorithm performance by applying it to a sequence of images of scenes subjected to abrupt illumination changes. Also a color based tracking is employed to show the importance of the color constancy algorithm in these scenes. Besides that, a color based visual-servo control working together with the color constancy algorithm is employed to guide a robot in an outdoor navigation task through an environment subjected to the variable illumination of the sun. The color constancy algorithm is also applied on images of an external environment that present illumination changes and the discussion of its utilization in place recognition, a fundamental task in robot localization, is made.