Multimedia content analysis for consumer electronics

The last few years have seen an explosion not only in the amount of content produced but also in the amount of content available. However, while it is great that users can watch virtually anything they want, it becomes increasingly difficult to find content that is actually relevant to the user. This "curse" of content availability is very important for consumer electronic devices since they are as valuable as the value they add to a user's life: if the only thing they do is to increase exponentially the users' choices of what to watch, listen to or consume in any way, then they add, paradoxically, more complication to their lives. Content analysis can be used to help the user select what to watch and is thus essential in modern multimedia devices. However, there are technical and human restrictions to the technology developed for such devices. In this paper we discuss these restrictions and how they affect the work on content analysis done in an industrial research environment. We will present three case studies that show how these restrictions influence the choices made from the moment content analysis technology is designed, until the moment it is tested with users.