ASLAN 2013 - IEEE International Workshop on Action Similarity in Unconstrained Videos

The goal of this workshop is to promote discussion and research into Action Recognition in unconstrained videos. This, by focusing on the following, atomic, "Action Pair-Matching" problem:
"Are actors, appearing in different, unconstrained videos, performing the same action?"
We focus on the "same vs not-same" question for action recognition, instead of category action labeling. The ability to compare two videos and correctly determine if they both present the same action requires fine-grained Action Recognition, and reasoning about Action Similarity. This is directly relevant to video retrieval, ranking, and similarity applications.
The ACTS'13 workshop is primarily designed as a performance evaluation / competition on the "The Action Similarity LAbeliNg" (ASLAN) benchmark. We solicit papers presenting new results on this benchmark and the workshop will recognize the best performing methods. In addition, following the central theme of Action Recognition in unconstrained, "in-the-wild" videos, we solicit also papers introducing new ideas and techniques related to Action Recognition in such conditions; presenting new results on the ASLAN set is recommended, but it is not a requirement for participation.