Spatiotemporal stacked sequential learning for Pedestrian detection

Alejandro González*, David Vázquez, Sebastian Ramos, Antonio M. López, Jaume Amores

*Autor corresponent d’aquest treball

Producció científica: Capítol de llibreContribució a congrés/conferènciaAvaluat per experts

4 Cites (Scopus)

Resum

Pedestrian classifiers decide which image windows contain a pedestrian. In practice, such classifiers provide a relatively high response at neighbor windows overlapping a pedestrian, while the responses around potential false positives are expected to be lower. An analogous reasoning applies for image sequences. If there is a pedestrian located within a frame, the same pedestrian is expected to appear close to the same location in neighbor frames. Therefore, such a location has chances of receiving high classification scores during several frames, while false positives are expected to be more spurious. In this paper we propose to exploit such correlations for improving the accuracy of base pedestrian classifiers. In particular, we propose to use two-stage classifiers which not only rely on the image descriptors required by the base classifiers but also on the response of such base classifiers in a given spatiotemporal neighborhood. More specifically, we train pedestrian classifiers using a stacked sequential learning (SSL) paradigm. We use a new pedestrian dataset we have acquired from a car to evaluate our proposal at different frame rates. We also test on well known dataset, Caltech. The obtained results show that our SSL proposal boosts detection accuracy significantly with a minimal impact on the computational cost. Interestingly, SSL improves more the accuracy at the most dangerous situations, i.e. when a pedestrian is close to the camera.

Idioma originalAnglès
Títol de la publicacióPattern Recognition and Image Analysis - 7th Iberian Conference, IbPRIA 2015, Proceedings
EditorsJaime S. Cardoso, Roberto Paredes, Xosé M. Pardo
EditorSpringer Verlag
Pàgines3-12
Nombre de pàgines10
ISBN (electrònic)9783319193892
DOIs
Estat de la publicacióPublicada - 2015
Publicat externament
Esdeveniment7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015 - Santiago de Compostela, Spain
Durada: 17 de juny 201519 de juny 2015

Sèrie de publicacions

NomLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volum9117
ISSN (imprès)0302-9743
ISSN (electrònic)1611-3349

Conferència

Conferència7th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2015
País/TerritoriSpain
CiutatSantiago de Compostela
Període17/06/1519/06/15

Fingerprint

Navegar pels temes de recerca de 'Spatiotemporal stacked sequential learning for Pedestrian detection'. Junts formen un fingerprint únic.

Com citar-ho