Ian Endres, Ming-Wei Chang, Vivek Srikumar, and Derek Hoiem
We cast the problem of recognizing related categories as a unified learning and structured prediction problem with shared body plans. When provided with detailed annotations of objects and their parts, these body plans model objects in terms of shared parts and layouts, simultaneously capturing a variety of categories in varied poses. We can use these body plans to jointly train many detectors in a shared framework with structured learning, leading to significant gains for each supervised task. Using our model, we can provide detailed predictions of objects and their parts for both familiar and unfamiliar categories.
|Published in||2012 IEEE Conference on Computer Vision and Pattern Recognition|
|Address||Los Alamitos, CA, USA|
|Publisher||IEEE Computer Society|