Temporally sorting images from real-world events

Autor(es)

Padilha, Rafael ^[1] ; Andalo, Fernanda A. ^[1] ; Lavi, Bahram ^[1] ; Pereira, Luis A. M. ^[1] ; Rocha, Anderson ^[1]

Afiliação do(s) autor(es)

^[1] Univ Estadual Campinas, Inst Comp, Campinas, SP - Brazil

• Tipo de documento: Artigo Científico
• Fonte: PATTERN RECOGNITION LETTERS; v. 147, p. 212-219, JUL 2021.
• Citações Web of Science: 2

Resumo

As smartphones become ubiquitous in modern life, every major event \& mdash; from musical concerts to ter-rorist attempts \& mdash; is massively captured by multiple devices and instantly uploaded to the Internet. Once shared through social media, the chronological order between available media pieces cannot be reliably recovered, hindering the understanding and reconstruction of that event. In this work, we propose data -driven methods for temporally sorting images originated from heterogeneous sources and captured from distinct angles, viewpoints, and moments. We model the chronological sorting task as an ensemble of binary classifiers whose answers are combined hierarchically to estimate an image \& rsquo;s temporal position within the duration of the event. We evaluate our method on images from the Notre-Dame Catedral fire and the Grenfell Tower fire events and discuss research challenges for analyzing data from real-world forensic events. Finally, we employ visualization techniques to understand what our models have learned, offering additional insights to the problem. (c) 2021 Elsevier B.V. All rights reserved. (AU)

• Processo FAPESP: 17/21957-2 - Encontrando Evidências Visuais da Passagem do Tempo
• Beneficiário: Rafael Soares Padilha
• Modalidade de apoio: Bolsas no Brasil - Doutorado
• Processo FAPESP: 18/16548-9 - Aprendendo pistas visuais da passagem do tempo
• Beneficiário: Luis Augusto Martins Pereira
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado
• Processo FAPESP: 18/05668-3 - Coerência espaço-temporal e de características a partir de dados heterogêneos
• Beneficiário: Bahram Lavi Sefidgari
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado
• Processo FAPESP: 17/12646-3 - Déjà vu: coerência temporal, espacial e de caracterização de dados heterogêneos para análise e interpretação de integridade
• Beneficiário: Anderson de Rezende Rocha
• Modalidade de apoio: Auxílio à Pesquisa - Temático