Update: Unfortunately, due to a low number of submissions, we are not able to guarantee a high scientific quality of the PIXSEM workshop and are therefore constrained to cancel the workshop. Successfully reviewed papers will be presented at the i-KNOW conference in the main track.
The 1st International Workshop “From Pixels to Semantics – Semantic Analysis meets Visual Analysis” will be held in conjunction with the International Conference on Knowledge Technologies and Data-driven Business (i-KNOW 2014), September 16-19, 2014, Graz, Austria.
With Flickr hosting more than 6 billion photos and Youtube users uploading more than 100 hours of video per minute, web-based image and video collections have become a number one source for many researchers seeking for large-scale multimedia datasets. While being most times freely available and the available content covering an extremely broad range of topics considering visual as well as semantic variance, datasets extracted from web communities have been used to evaluate multimedia retrieval and data mining algorithms, for explorative search and retrieval, and also as training sets for content-based classification methods. Typically, multimedia data from web-based platforms comes with user generated text annotations such as descriptions, user comments and tags providing user generated links between the visual and semantic domain – a kind of metadata which is usually very expensive to create. However, this advantage comes with a cost in terms of a loose linkage between the semantics of the metadata and the semantics of the depicted visual information which must not necessarily correspond in all cases. Taking into account both visual information as well as user generated text-based metadata enables new ways to combine visual analysis with semantic technologies to bridge the gap between mere statistical analysis and image understanding.
This workshop gathers researchers working in domains related to the Semantic Web and Computer Vision community such as: semantic web technologies, artificial intelligence, information retrieval, multimedia, and communication technologies, social media mining, knowledge mining, data science, human-computer interaction, humanities, and web information systems. We highly encourage the presentation of work that targets the intersection of both fields. We expect the participants to present novel ideas on how semantic analysis may help to alleviate classification and retrieval problems in computer vision and how algorithms from computer vision may be exploited to address additional layers of semantics in multimedia documents.