Interactive Adaptation of Real-Time Object Detectors

TitleInteractive Adaptation of Real-Time Object Detectors
Publication TypeConference Paper
Year of Publication2014
AuthorsGöhring, D., Hoffman J., Rodner E., Saenko K., & Darrell T.
Other Numbers3630
Abstract

In the following paper, we present a framework forquickly training 2D object detectors for robotic perception. Ourmethod can be used by robotics practitioners to quickly (under30 seconds per object) build a large-scale real-time perceptionsystem. In particular, we show how to create new detectors onthe fly using large-scale internet image databases, thus allowinga user to choose among thousands of available categories tobuild a detection system suitable for the particular roboticapplication. Furthermore, we show how to adapt these modelsto the current environment with just a few in-situ images.Experiments on existing 2D benchmarks evaluate the speed,accuracy, and flexibility of our system.

URLhttps://www.icsi.berkeley.edu/pubs/vision/interactiveadaption14.pdf
Bibliographic Notes

Proceedings of the IEEE International Conference in Robotics and Automation (ICRA), Hong Kong, China

Abbreviated Authors

D. Goehring, J. Hoffman, E. Rodner, K. Saenko, and T. Darrell

ICSI Research Group

Vision

ICSI Publication Type

Article in conference proceedings