The second official REPERE evaluation

The REPERE Challenge aims at supporting research on people recognition in multimodal conditions. Following a 2012 dry-run [1] and a first official campaign [2], the second official evaluation of systems has been conducted at the beginning of 2014. To both help system development and assess the technology progress a specific corpus has been developed. It current totals 60 hours of video with multimodal annotations. The systems have to answer the following questions: Who is speaking? Who is present in the video? What names are cited? What names are displayed? The challenge is to combine the various informations coming from the speech and the images. For the 2014 edition, a surprise show has been added to the test corpus and a new method to estimate the significance of the performance difference has been introduced.