Videntifier" Forensic: large-scale video identification in practice

Authors:
Herwig Lejsek;Hrönn Þormóðsdóttir;Friðrik Ásmundsson;Kristleifur Daðason;Ársæll Þór Jóhannsson;Björn Þór Jónsson;Laurent Amsaleg
Affiliations:
Videntifier Technologies ehf, and Reykjavík University, Reykjavík, Iceland;Videntifier Technologies ehf, Reykjavík, Iceland;Videntifier Technologies ehf, Reykjavík, Iceland;Videntifier Technologies ehf, Reykjavík, Iceland;Videntifier Technologies ehf, Reykjavík, Iceland;Reykjavík University, Reykjavík, Iceland;IRISA-CNRS, Rennes, France
Venue:
Proceedings of the 2nd ACM workshop on Multimedia in forensics, security and intelligence
Year:
2010

Citing 5
Cited 3

Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
NV-Tree: An Efficient Disk-Based Index for Approximate Search in Very Large High-Dimensional Collections

IEEE Transactions on Pattern Analysis and Machine Intelligence
Videntifier™ forensic: a new law enforcement service for automatic identification of illegal video material

MiFor '09 Proceedings of the First ACM workshop on Multimedia in forensics
Videntifier™ forensic: robust and efficient detection of illegal multimedia

MM '09 Proceedings of the 17th ACM international conference on Multimedia
GPU acceleration of Eff2 descriptors using CUDA

Proceedings of the international conference on Multimedia

Second ACM international workshop on multimedia in forensics, security and intelligence (MiFor 2010)

Proceedings of the international conference on Multimedia
NV-Tree: nearest neighbors at the billion scale

Proceedings of the 1st ACM International Conference on Multimedia Retrieval
A multimedia analytics framework for browsing image collections in digital forensics

Proceedings of the 20th ACM international conference on Multimedia

Quantified Score

Hi-index	0.00

Visualization

Abstract

Identifying videos on seized hard drives and other storage devices is a very tedious and time consuming task for forensic investigators. In particular, the vast amount of available material on the Internet and the large storage capacities of today's hard drives have become a strong headache for them. Videntifier" Forensic is a recent service for forensic video identification, which is based on state-of-the-art high-dimensional descriptors and high-dimensional indexing. In this paper we describe how Videntifier" Forensic tackles very large collections of video material and how robust it is towards standard modifications. We then present measurements that involve four different datasets and three collection sizes of up to 25,000 hours of video content. Our results show that Videntifier" Forensic scales very well, both in terms of the efficiency and effectiveness of the service.