Fast appearance based object recognition: A hybrid approach

Philip Blackwell; David Austin

doi:10.1109/ROBOT.2005.1570110

Fast appearance based object recognition: A hybrid approach

Philip Blackwell^*, David Austin

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Visual object recognition is a useful skill for robots to possess. However, present approaches to the problem do not scale to large numbers of objects (few manage more than 10) and require too much computation for real-time tasks on a robot. This paper presents a hybrid decision tree/support vector machine approach to recognition which is fast, with recognition times under one second. A new test dataset is also presented, consisting of over 100,000 images of Lego bricks, acquired by repeatedly dropping the bricks. The proposed method achieves 96% accuracy on the set of 89 different types of Lego bricks, demonstrating its applicability for large-scale real-time visual object recognition.

Original language	English
Title of host publication	Proceedings of the 2005 IEEE International Conference on Robotics and Automation
Pages	144-149
Number of pages	6
DOIs	https://doi.org/10.1109/ROBOT.2005.1570110
Publication status	Published - 2005
Event	2005 IEEE International Conference on Robotics and Automation - Barcelona, Spain Duration: 18 Apr 2005 → 22 Apr 2005

Publication series

Name	Proceedings - IEEE International Conference on Robotics and Automation
Volume	2005
ISSN (Print)	1050-4729

Conference

Conference	2005 IEEE International Conference on Robotics and Automation
Country/Territory	Spain
City	Barcelona
Period	18/04/05 → 22/04/05

Access to Document

10.1109/ROBOT.2005.1570110

Cite this

@inproceedings{80d905a7621d4a04969e57651c98eb06,

title = "Fast appearance based object recognition: A hybrid approach",

abstract = "Visual object recognition is a useful skill for robots to possess. However, present approaches to the problem do not scale to large numbers of objects (few manage more than 10) and require too much computation for real-time tasks on a robot. This paper presents a hybrid decision tree/support vector machine approach to recognition which is fast, with recognition times under one second. A new test dataset is also presented, consisting of over 100,000 images of Lego bricks, acquired by repeatedly dropping the bricks. The proposed method achieves 96% accuracy on the set of 89 different types of Lego bricks, demonstrating its applicability for large-scale real-time visual object recognition.",

keywords = "Decision trees, Large database, Object recognition",

author = "Philip Blackwell and David Austin",

year = "2005",

doi = "10.1109/ROBOT.2005.1570110",

language = "English",

isbn = "078038914X",

series = "Proceedings - IEEE International Conference on Robotics and Automation",

pages = "144--149",

booktitle = "Proceedings of the 2005 IEEE International Conference on Robotics and Automation",

note = "2005 IEEE International Conference on Robotics and Automation ; Conference date: 18-04-2005 Through 22-04-2005",

}

Blackwell, P & Austin, D 2005, Fast appearance based object recognition: A hybrid approach. in Proceedings of the 2005 IEEE International Conference on Robotics and Automation., 1570110, Proceedings - IEEE International Conference on Robotics and Automation, vol. 2005, pp. 144-149, 2005 IEEE International Conference on Robotics and Automation, Barcelona, Spain, 18/04/05. https://doi.org/10.1109/ROBOT.2005.1570110

Fast appearance based object recognition: A hybrid approach. / Blackwell, Philip; Austin, David.
Proceedings of the 2005 IEEE International Conference on Robotics and Automation. 2005. p. 144-149 1570110 (Proceedings - IEEE International Conference on Robotics and Automation; Vol. 2005).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Fast appearance based object recognition

T2 - 2005 IEEE International Conference on Robotics and Automation

AU - Blackwell, Philip

AU - Austin, David

PY - 2005

Y1 - 2005

N2 - Visual object recognition is a useful skill for robots to possess. However, present approaches to the problem do not scale to large numbers of objects (few manage more than 10) and require too much computation for real-time tasks on a robot. This paper presents a hybrid decision tree/support vector machine approach to recognition which is fast, with recognition times under one second. A new test dataset is also presented, consisting of over 100,000 images of Lego bricks, acquired by repeatedly dropping the bricks. The proposed method achieves 96% accuracy on the set of 89 different types of Lego bricks, demonstrating its applicability for large-scale real-time visual object recognition.

AB - Visual object recognition is a useful skill for robots to possess. However, present approaches to the problem do not scale to large numbers of objects (few manage more than 10) and require too much computation for real-time tasks on a robot. This paper presents a hybrid decision tree/support vector machine approach to recognition which is fast, with recognition times under one second. A new test dataset is also presented, consisting of over 100,000 images of Lego bricks, acquired by repeatedly dropping the bricks. The proposed method achieves 96% accuracy on the set of 89 different types of Lego bricks, demonstrating its applicability for large-scale real-time visual object recognition.

KW - Decision trees

KW - Large database

KW - Object recognition

UR - http://www.scopus.com/inward/record.url?scp=33846167779&partnerID=8YFLogxK

U2 - 10.1109/ROBOT.2005.1570110

DO - 10.1109/ROBOT.2005.1570110

M3 - Conference contribution

SN - 078038914X

SN - 9780780389144

T3 - Proceedings - IEEE International Conference on Robotics and Automation

SP - 144

EP - 149

BT - Proceedings of the 2005 IEEE International Conference on Robotics and Automation

Y2 - 18 April 2005 through 22 April 2005

ER -

Fast appearance based object recognition: A hybrid approach

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this