Szczegóły publikacji
Opis bibliograficzny
Enhanced method of near duplicate detection for red carpet photographs / Michał GREGA // W: Multimedia Communications, Services and Security : 8th international conference, MCSS 2015 : Krakow, Poland, November 24, 2015 : proceedings / eds. Andrzej Dziech, Mikołaj Leszczuk, Remigiusz Baran. — Cham [etc.] : Springer International Publishing, cop. 2015. — (Communications in Computer and Information Science ; ISSN 1865-0929 ; 566). — ISBN: 978-3-319-26403-5; e-ISBN: 978-3-319-26404-2. — S. 132–140. — Bibliogr. s. 139–140, Abstr.
Autor
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 94540 |
|---|---|
| Data dodania do BaDAP | 2016-01-05 |
| DOI | 10.1007/978-3-319-26404-2_11 |
| Rok publikacji | 2015 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp | |
| Konferencja | Multimedia Communications, Services and Security 2015 |
| Czasopismo/seria | Communications in Computer and Information Science |
Abstract
The DEEP service is comprehensive new content discovery solution that offers fun, creative, modern way to discover information. Automated generation of content for DEEP relies on structured and unstructured sources of data and on multimedia databases. Unfortunately, using the Internet as a source for multimedia can result in the acquisition of so-called "Near Duplicates" - visually similar images. In this paper we propose an enhanced method of Near Duplicate detection for a special kind of photographs - images of celebrities commonly known as "red carpet" photographs. We have observed that near duplicates of such photos are most commonly crops of busts (head and upper torso) of the celebrity. We have combined an automated method for bust cropping with similarity computation and have obtained 95% sensitivity and 99% specificity for near duplicate detection.