The Orange Labs Real Time Video Copy Detection System - TrecVid 2008 Results

In this paper, we describe the content-based video copy detection system developed at Orange Labs. We also present the evaluation results of the TrecVid 2008 copy detection evaluation task. Target applications of video copy detection are mainly related to video copyright protection and video database management (e.g. duplicate detection). We have taken into account the constraints of these applications when building our system. We have in particular focused on the objective of achieving an optimal trade-off between effectiveness (i.e. robustness to transformations) and efficiency (i.e. rapidity). The 3 runs we have submitted to TrecVid 2008 are described and analyzed according to the evaluation criteria defined for the benchmark. The obtained results show that our goal of having an efficient and effective system has been globally reached.

• Type: Conference
• Year: 2008
• Where: TRECVID
• Authors: Nicolas Gengembre, Sid-Ahmed Berrani
• Keywords: Copy Detection, Copy Detection Evaluation, Internet Technology, TRECVID 2008, Video Copy Detection