Closed-loop object recognition using reinforcement learning

Jing Peng, Bir Bhanu

Research output: Contribution to journalArticleResearchpeer-review

82 Citations (Scopus)

Abstract

Current computer vision systems whose basic methodology is open-loop or fi'/tertype typically use image segmentation followed by object recognition algorithms. These systems are not robust for most real-world applications. In contrast the system presented here achieves robust performance by using reinforcement learning to induce a mapping from input images to corresponding segmentation parameters. This is accomplished by using the confidence level of model matching as a reinforcement signal for a team of learning automata to search for segmentation parameters during training. The use of the recognition algorithm as part of the evaluation function for image segmentation gives rise to significant improvement of the system performance by automatic generation of recognition strategies. The system is verified through experiments on sequences of indoor and outdoor color images with varying external conditions.

Original languageEnglish
Pages (from-to)139-154
Number of pages16
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume20
Issue number2
DOIs
StatePublished - 1 Dec 1998

Fingerprint

Object recognition
Reinforcement learning
Object Recognition
Reinforcement Learning
Image segmentation
Closed-loop
Recognition Algorithm
Image Segmentation
Function evaluation
Segmentation
Computer vision
Learning Automata
Model Matching
Reinforcement
Robust Performance
Confidence Level
Evaluation Function
Vision System
Color Image
Real-world Applications

Keywords

  • Adaptive color image segmentation
  • Function optimization
  • Generalized learning automata
  • Learning in computer vision
  • Model-based object recognition
  • Multiscenario recognition
  • Parameter learning
  • Recognition feedback
  • Segmentation evaluation

Cite this

@article{a38da8d0cc4e4d3683396f53e3c43313,
title = "Closed-loop object recognition using reinforcement learning",
abstract = "Current computer vision systems whose basic methodology is open-loop or fi'/tertype typically use image segmentation followed by object recognition algorithms. These systems are not robust for most real-world applications. In contrast the system presented here achieves robust performance by using reinforcement learning to induce a mapping from input images to corresponding segmentation parameters. This is accomplished by using the confidence level of model matching as a reinforcement signal for a team of learning automata to search for segmentation parameters during training. The use of the recognition algorithm as part of the evaluation function for image segmentation gives rise to significant improvement of the system performance by automatic generation of recognition strategies. The system is verified through experiments on sequences of indoor and outdoor color images with varying external conditions.",
keywords = "Adaptive color image segmentation, Function optimization, Generalized learning automata, Learning in computer vision, Model-based object recognition, Multiscenario recognition, Parameter learning, Recognition feedback, Segmentation evaluation",
author = "Jing Peng and Bir Bhanu",
year = "1998",
month = "12",
day = "1",
doi = "10.1109/34.659932",
language = "English",
volume = "20",
pages = "139--154",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",
number = "2",

}

Closed-loop object recognition using reinforcement learning. / Peng, Jing; Bhanu, Bir.

In: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 20, No. 2, 01.12.1998, p. 139-154.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Closed-loop object recognition using reinforcement learning

AU - Peng, Jing

AU - Bhanu, Bir

PY - 1998/12/1

Y1 - 1998/12/1

N2 - Current computer vision systems whose basic methodology is open-loop or fi'/tertype typically use image segmentation followed by object recognition algorithms. These systems are not robust for most real-world applications. In contrast the system presented here achieves robust performance by using reinforcement learning to induce a mapping from input images to corresponding segmentation parameters. This is accomplished by using the confidence level of model matching as a reinforcement signal for a team of learning automata to search for segmentation parameters during training. The use of the recognition algorithm as part of the evaluation function for image segmentation gives rise to significant improvement of the system performance by automatic generation of recognition strategies. The system is verified through experiments on sequences of indoor and outdoor color images with varying external conditions.

AB - Current computer vision systems whose basic methodology is open-loop or fi'/tertype typically use image segmentation followed by object recognition algorithms. These systems are not robust for most real-world applications. In contrast the system presented here achieves robust performance by using reinforcement learning to induce a mapping from input images to corresponding segmentation parameters. This is accomplished by using the confidence level of model matching as a reinforcement signal for a team of learning automata to search for segmentation parameters during training. The use of the recognition algorithm as part of the evaluation function for image segmentation gives rise to significant improvement of the system performance by automatic generation of recognition strategies. The system is verified through experiments on sequences of indoor and outdoor color images with varying external conditions.

KW - Adaptive color image segmentation

KW - Function optimization

KW - Generalized learning automata

KW - Learning in computer vision

KW - Model-based object recognition

KW - Multiscenario recognition

KW - Parameter learning

KW - Recognition feedback

KW - Segmentation evaluation

UR - http://www.scopus.com/inward/record.url?scp=0031998786&partnerID=8YFLogxK

U2 - 10.1109/34.659932

DO - 10.1109/34.659932

M3 - Article

VL - 20

SP - 139

EP - 154

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 2

ER -