Parametrized Shape Models for Clothing

TitleParametrized Shape Models for Clothing
Publication TypeConference Paper
Year of Publication2011
AuthorsMiller, S., Fritz M., Darrell T., & Abbeel P.
Other Numbers3213
Abstract

We consider the problem of recognizing the configurationof clothing articles when crudely spread out on aflat surface, prior to and during folding. At the core of ourapproach are parametrized shape models for clothing articles.Each clothing category has its own shape model, and the varietyin shapes for a given category is achieved through variation ofthe parameters. We present an efficient algorithm to find theparameters that provide the best fit when given an image of aclothing article. The models are such that, once the parametershave been fit, they provide a basic parse of the clothing article,allowing it to be followed by autonomous folding from categorylevel specifications of fold sequences. Our approach is also ableto recover the configuration of a clothing article when folds arebeing introduced—an important feature towards closing theperception-action loop. Additionally, our approach provides areliable method of shape-based classification, simply by examiningwhich model yields the best fit. Our experiments illustratethe effectiveness of our approach on a large set of clothingarticles. Furthermore, we present an end-to-end system, whichstarts from an unknown spread-out clothing article, performsa parametrized model fit, then follows a category-level (ratherthan article specific) set of folding instructions, closing the loopwith perceptual feedback by re-fitting between folds.

Acknowledgment

This work was supported in part by NSF under award IIS-0904672, Willow Garage under the PR2 Beta Program, anda Feodor Lynen Fellowship granted by the Alexander vonHumboldt Foundation.

URLhttp://www.icsi.berkeley.edu/pubs/vision/parametrizedshape11.pdf
Bibliographic Notes

Proceedings of the 2011 IEEE International Conference on Robotics and Automation (ICRA 2011), Shanghai, China

Abbreviated Authors

S. Miller, M. Fritz, T. Darrell, and P. Abbeel

ICSI Research Group

Vision

ICSI Publication Type

Article in conference proceedings