Portal de Periódicos da CAPES

Efficient Search and Browsing of Large-Scale Video Collections with Vibro

2022; Springer Science+Business Media; Linguagem: Inglês

10.1007/978-3-030-98355-0_43

1611-3349

Nico Hezel, Konstantin Schall, Klaus Jung, Kai Uwe Barthel,

Multimodal Machine Learning Applications

In this paper, we present the newest version of our interactive video browser tool Vibro. For this iteration, we focused on improving the user interface to enable a more accessible temporal search, upgrading the shot-detection algorithm, replacing a keyword-based search with rich text input, and reducing query times by applying a graph-based approximate nearest neighbor search method. With these extensive updates, we feel well-equipped to handle the huge amounts of data coming our way in the next VBS competitions and achieve competitive results in the contest.