Abstract: We propose a visual event recognition framework for consumer domain videos by leveraging a large amount of loosely labeled web videos (e.g., from YouTube). First, we propose a new aligned ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results