'Workshop on higher-order models and global constraints in computer vision', Florence, Italy
Many tasks in computer vision have been modelled as discrete labelling problems. However, this method severely restricts the representational power of these models as they are unable to capture the richness of natural images. New methods are making it possible to encode significantly more sophisticated priors and structural dependencies between variables, making images more vivid and rich.
The goal of this workshop will be to bring together researchers working on different aspects of the topic (such as in modelling, inference and learning) and discuss various techniques, common solutions, open questions and future pursuits.For further information, please visit: http://sites.google.com/a/ttic.edu/eccv-2012-workshop-hipot/