Coarse-to-Fine Adaptive People Detection for Video Sequences by Maximizing Mutual Information

Biblos-e Archivo/Manakin Repository

Show simple item record

dc.contributor.author García-Martín, Álvaro es_ES
dc.contributor.author Sanmiguel, Juan Carlos es_ES
dc.contributor.author Martínez, José M. es_ES
dc.contributor.other UAM. Departamento de Tecnología Electrónica y de las Comunicaciones es_ES
dc.date.accessioned 2020-01-28T16:29:35Z es_ES
dc.date.available 2020-01-28T16:29:35Z es_ES
dc.date.issued 2018-12-20 es_ES
dc.identifier.citation Sensors 19.4(2019): 1-22 en_US
dc.identifier.issn 1424-8220 es_ES
dc.identifier.uri http://hdl.handle.net/10486/689969 en_US
dc.description.abstract Applying people detectors to unseen data is challenging since patterns distributions, such as viewpoints, motion, poses, backgrounds, occlusions and people sizes, may significantly differ from the ones of the training dataset. In this paper, we propose a coarse-to-fine framework to adapt frame by frame people detectors during runtime classification, without requiring any additional manually labeled ground truth apart from the offline training of the detection model. Such adaptation make use of multiple detectors mutual information, i.e., similarities and dissimilarities of detectors estimated and agreed by pair-wise correlating their outputs. Globally, the proposed adaptation discriminates between relevant instants in a video sequence, i.e., identifies the representative frames for an adaptation of the system. Locally, the proposed adaptation identifies the best configuration (i.e., detection threshold) of each detector under analysis, maximizing the mutual information to obtain the detection threshold of each detector. The proposed coarse-to-fine approach does not require training the detectors for each new scenario and uses standard people detector outputs, i.e., bounding boxes. The experimental results demonstrate that the proposed approach outperforms state-of-the-art detectors whose optimal threshold configurations are previously determined and fixed from offline training data en_US
dc.description.sponsorship This work has been partially supported by the Spanish government under the project TEC2014-53176-R (HAVideo) en_US
dc.format.extent 22 pág. es_ES
dc.format.mimetype application/pdf en_US
dc.language.iso eng en_US
dc.publisher MDPI en_US
dc.relation.ispartof Sensors en_US
dc.rights © 2018 by the authors en_US
dc.subject.other People detection en_US
dc.subject.other Detector adaptation en_US
dc.subject.other Pair-wise correlation en_US
dc.title Coarse-to-Fine Adaptive People Detection for Video Sequences by Maximizing Mutual Information en_US
dc.type article en_US
dc.subject.eciencia Telecomunicaciones es_ES
dc.relation.publisherversion https://doi.org/10.3390/s19010004 en_US
dc.identifier.doi 10.3390/s19010004 es_ES
dc.identifier.publicationfirstpage 1 es_ES
dc.identifier.publicationissue 4 es_ES
dc.identifier.publicationlastpage 22 es_ES
dc.identifier.publicationvolume 19 es_ES
dc.relation.projectID Gobierno de España. TEC2014-53176-R es_ES
dc.type.version info:eu-repo/semantics/publishedVersion en_US
dc.contributor.group Tratamiento e Interpretación de Vídeo (ING EPS-06) es_ES
dc.rights.accessRights openAccess en_US
dc.authorUAM García Martín, Álvaro (261833) es_ES
dc.authorUAM San Miguel Avedillo, Juan Carlos (261249) es_ES
dc.authorUAM Martínez Sánchez, José María (260488) es_ES


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record