Actas de congresos
Temporal-and spatial-driven video summarization using optimum-path forest
Proceedings - 2016 29th SIBGRAPI Conference on Graphics, Patterns and Images, SIBGRAPI 2016, p. 335-339.
Universidade Estadual Paulista (UNESP)
Universidade de São Paulo (USP)
Video summarization aims at generating reduced representations for fast and effective video retrieval and classification. In this paper, we cope with such problem by proposing a temporal-and spatial-driven approach that makes use of the Optimum-Path Forest (OPF) clustering to automatic find the number of keyframes, as well as to extract them to compose the final summary. The experiments in two public datasets show OPF can outperform very recent results, thus achieving a performance comparable to some state-of-the-art techniques.