Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ Estudo Geralarrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
Estudo Geral
Master thesis . 2021
Data sources: Estudo Geral
addClaim

This Research product is the result of merged Research products in OpenAIRE.

You have already added 0 works in your ORCID record related to the merged Research product.

Computer Vision

Authors: Santos, Francisco José Marques;

Computer Vision

Abstract

The last few years have brought a growth in the usage of technologies like Deep Learning and Convolution Neural Networks, that is possible due to advances in hardware and related areas. In reality, the technology that exists today allows anyone to work with neural networks with low ranked Graphics Processing Unit (GPU). Two different networks were exploited, more specifically RCNN and End-to-end Network. Both were trained to get speed estimation of vehicles. Concerning NN, the main propose was to detect objects in an image to feed a tracker, after obtaining the calibration parameters through an automatic calibration method developed in ISR. On this system, the mean error achieved for the speed estimation was 3.64km/h.In the End-to-End network, the Deep learning component is even more notorious. This network has the capacity of estimating velocity without resorting to auxiliary systems such as calibration and tracking. In this document, the network presented is used in speed estimation with minimal intervention. This network has achieved a mean speed estimation error between 3km/h and 5km/h. Both Networks can estimate vehicles speed with satisfactory precision. This project was in the interest of Brisa, a toll collection service, aiming at the use of surveillance cameras capable of monitoring traffic and at the same time detecting speed through neural networks without the need to give pre-calculated information to the network. .

Nos ́ultimos anos houve um aumento no uso de tecnologias como Deep Learning e Convolutional Neural Network (CNN), isto foi possivel devido ao avanço no hardware necessário e nas ́areas relacionadas. Hoje em dia, a tecnologia existente permite que qualquer pessoa possa utilizar redes neuronais com um GPU médio. Neste trabalho, explorou-se duas redes diferentes, mais concretamente RCNN e End-to-end Network. Ambas foram treinadas de forma a estimar a velocidade de veiculos. Relativamente à RCNN, o objetivo foi detetar objetos numa imagem deforma a alimentar um tracker depois de obtidos os parâmetros de calibração, através de um método de calibração automático desenvolvido no ISR. Na rede End-to-end, a componente de Deep learning ́e notória pois consegue estimar a velocidade sem necessidade de sitemas auxiliares como calibração e tracking. É possivel observar que esta rede tem grandes potencialidades na estimação de velocidades com o mínimo de intervenção. Ambas as redes conseguem estimar a velocidade de veículos com uma precisão satisfatória. Este projeto foi do interesse da Brisa, um serviço de cobrança de portagens, tendo como objectivo utilizar câmaras de vigilância com capacidade de monitoriação de tráfego e ao mesmo tempo detetar velocidades através de redes neuronais, sem a necessidade de fornecer informação calculada a priori à rede. .

Dissertação de Mestrado Integrado em Engenharia Electrotécnica e de Computadores apresentada à Faculdade de Ciências e Tecnologia

Related Organizations
Keywords

End-to-end Network, Vehicle Recognition, Deep Learning, Computer Vision, Convolutional Neural Networks, Reconhecimento de Veículos, Visão por Computador

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Green