36. SIMULTANEOUS LEARNING AND ALIGNMENT: MULTI-INSTANCE AND MULTI-POSE LEARNING
Department: Computer Science & Engineering
Faculty Advisor(s):
Serge Belongie
Primary Student
Name: Boris Babenko
Email: bbabenko@ucsd.edu
Phone: 858-534-8187
Grad Year: 2011
Abstract
In object recognition in general and in face detection in particular, data alignment is necessary to achieve good classification results with certain statistical learning approaches such as Viola-Jones. Data can be aligned in one of two ways: (1) by separating the data into coherent groups and training separate classifiers for each; (2) by adjusting training samples so they lie in correspondence. If done manually, both procedures are labor intensive and can significantly add to the cost of labeling. In this paper we present a unified boosting framework for simultaneous learning and alignment. We present a novel boosting algorithm for Multiple Pose Learning (MPL), where the goal is to simultaneously split data into groups and train classifiers for each. We also review Multiple Instance Learning (MIL), and in particular mil-boost, and describe how to use it to simultaneously train a classifier and bring data into correspondence. We show results on variations of LFW and MNIST, demonstrating the potential of these approaches. These results are preliminary but promising. Significant testing remains to evaluate if these approaches can improve state of the art detection systems. Finally, we describe a number of interesting extensions to MPL and MIL that warrant further investigation. We hope to pursue these extensions in future work.
Related Links:
All Content Copyright 2008 Regents of the University of California. All rights reserved.
Official Web page of the University of California, San Diego.
