Object detection and association in multiview scenarios based on Deep learning

Biblos-e Archivo/Manakin Repository

Show simple item record

dc.contributor.advisor García Martín, Álvaro (director)
dc.contributor.advisor Sanmiguel, Juan Carlos (supervisor)
dc.contributor.author Moral de Eusebio, Paula
dc.contributor.other UAM. Departamento de Tecnología Electrónica y de las Comunicaciones es_ES
dc.date.accessioned 2019-09-11T15:51:25Z
dc.date.available 2019-09-11T15:51:25Z
dc.date.issued 2019-07
dc.identifier.uri http://hdl.handle.net/10486/688569 en_US
dc.description Máster Universitario Image Processing and Computer Vision en_US
dc.description.abstract The detection and association of objects in multiview scenarios is an area of research within the Computer Vision that is very useful in tasks such as video surveillance, for example when identifying in di erent scenes a person who has carried out any kind of anomaly. This project is going to focus on vehicle re-identification, due to it has been a critical problem in the Intelligent Transportation System (ITS) for the recent years, and we can see our performance compared with the state of the art participating in the 2019 NVIDIA AI City Challenge. The main objective of this Master Thesis is the development of a system that detects and associates multiple objects in multiview scenarios based on deep learning. For this task, different algorithms from the state of the art have been studied. Furthermore, the method implemented uses feature extraction methods and metric learning techniques and it returns a list with all the matches between the query object and the images from the gallery set, sorted according to their distance. A new dataset has been reorganized from the train part of the CityFlow-ReID dataset. In order to improve the results, it is included a metric network combination at distances and ranks level from the different feature extraction methods and metric learning techniques. Finally, we have developed our own experimental setup. en_US
dc.format.extent 75 pág. es_ES
dc.format.mimetype application/pdf en_US
dc.language.iso eng en_US
dc.subject.other Vehicle re-identification en_US
dc.subject.other features extraction en_US
dc.subject.other metric learning en_US
dc.title Object detection and association in multiview scenarios based on Deep learning en_US
dc.type masterThesis en_US
dc.subject.eciencia Telecomunicaciones es_ES
dc.rights.cc Reconocimiento – NoComercial – SinObraDerivada es_ES
dc.rights.accessRights openAccess en_US

Files in this item


This item appears in the following Collection(s)

Show simple item record