Scalability of the NV-tree: Three Experiments

Laurent Amsaleg (1) , Björn Thor Jónsson (2) , Herwig Lejsek (3)

1 LinkMedia - Creating and exploiting explicit links between multimedia fragments
2 ITU - IT University of Copenhagen
3 Videntifier Technologies

The NV-tree is a scalable approximate high-dimensional indexing method specifically designed for large-scale visual instance search. In this paper, we report on three experiments designed to evaluate the performance of the NV-tree. Two of these experiments embed standard benchmarks within collections of up to 28.5 billion features, representing the largest single-server collection ever reported in the literature. The results show that indeed the NV-tree performs very well for visual instance search applications over large-scale collections.

Domaines: Informatique [cs]; Multimédia [cs.MM]

Laurent Amsaleg, Björn Thor Jónsson, Herwig Lejsek. Scalability of the NV-tree: Three Experiments. SISAP 2018 - 11th International Conference on Similarity Search and Applications, Oct 2018, Lima, Peru. pp.1-14. ⟨hal-01843046⟩