Techniques for Effective and Efficient Fire Detection from Social Media Images

Abstract

Abstract:Social media could provide valuable information to support decision making in crisis management, such as in accidents, explosions and fires. However, much of the data from social media are images, which are uploaded in a rate that makes it impossible for human beings to analyze them. Despite the many works on image analysis, there are no fire detection studies on social media. To fill this gap, we propose the use and evaluation of a broad set of content-based image retrieval and classification techniques for fire detection. Our main contributions are: (i) the development of the Fast-Fire Detection method (FFireDt), which combines feature extractor and evaluation functions to support instance-based learning; (ii) the construction of an annotated set of images with ground-truth depicting fire occurrences -the Flickr-Fire dataset; and (iii) the evaluation of 36 efficient image descriptors for fire detection. Using real data from Flickr, our results showed that FFireDt was able to achieve a precision for fire detection comparable to that of human annotators. Therefore, our work shall provide a solid basis for further developments on monitoring images from social media.

Conquering knowledge from images: improving image mining with region-based analysis and associated information

“…In practice, different FEMs can be combined to describe various characteristics and improve the content representation being analyzed. Examples of works in this line of research are (CHAUDHRY et al, 2009;CHINO et al, 2018;BEDO et al, 2015;. After extracting the relevant feature vectors with FEMs, it is necessary to establish a metric that allows the comparison of pairs of such vectors.…”

“…Many classifiers have shown to be fit for the image classification task in the literature. Example of recent works are (BEDO et al, 2015;KALE et al, 2016;KARARGYRIS et al, 2016;CHINO et al, 2018). Usually, they rely on color, texture, and other available features to perform classification.…”

Indexação Eficiente de Múltiplos Espaços Métricos para Otimização de Consultas por Similaridade

“…Por exemplo, considere um cenário em que um usuário deseja consultar imagens explorando recursos de cor e textura. Muitas vezes, os vetores que representam cor e textura são concatenados, gerando um único vetor resultante, que é indexado por um MAM (BEDO et al, 2015). No entanto, eles permitem responder apenas a perguntas considerando todos os recursos ao mesmo tempo, e traz à tona o problema da chamada maldição da alta dimensionalidade (BUSTOS; KEIM; SCHRECK, 2005), pois trabalhar com dados de alta dimensionalidade geralmente prejudica a representação dos objetos e compromete o desempenho das funções de recuperação.…”