Projects per year
Videos are often associated with additional information that could be valuable for interpretation of their content. This especially applies for the recognition of faces within video streams, where often cues such as transcripts and subtitles are available. However, this data is not completely reliable and might be ambiguously labeled. To overcome these limitations, we take advantage of semi-supervised (SSL) and multiple instance learning (MIL) and propose a new semi-supervised multiple instance learning (SSMIL) algorithm. Thus, during training we can weaken the prerequisite of knowing the label for each instance and can integrate unlabeled data, given only probabilistic information in form of priors. The benefits of the approach are demonstrated for face recognition in videos on a publicly available benchmark dataset. In fact, we show exploring new information sources can considerably improve the classification results.
|Title of host publication||2011 8th IEEE International Conference on Advanced Video and Signal Based Surveillance, AVSS 2011|
|Pages||23 - 28|
|Publication status||Published - 2011|
|Event||IEEE International Conference on Advanced Video and Signal Based Surveillance: AVSS 2011 - Klagenfurt, Austria|
Duration: 30 Aug 2011 → 2 Sept 2011
|Conference||IEEE International Conference on Advanced Video and Signal Based Surveillance|
|Period||30/08/11 → 2/09/11|
Fields of Expertise
- Information, Communication & Computing
- 1 Finished
MDL - Multimedia Documentation Lab: Wissensrepäsentation und -Organisation bei multimedialen Inhalten als Voraussetzung für sicherheitsrelevante Analysen
Wohlhart, P., Köstinger, M. & Bischof, H.
1/02/09 → 31/01/12
Project: Research project