Inducing a perceptual relevance shape classifier

Victoria J. Hodge; John P. Eakins; James Austin

doi:10.1145/1282280.1282306

Inducing a perceptual relevance shape classifier

Victoria J. Hodge, John P. Eakins, James Austin

Computer Science

Research output: Contribution to conference › Paper › peer-review

Abstract

In this paper, we develop a system to classify the outputs of image segmentation algorithms as perceptually relevant or perceptually irrelevant with respect to human perception. The work is aimed at figurative images. We previously investigated human visual perception of trademark images and established a body of ground truth data in the form of trademark images and their respective human segmentations. The work indicated that there is a core set of segmentations for each image that people perceive. Here we use this core set of segmentations to train a classifier to classify closed shapes output from an image segmentation algorithm so that the method returns the image segments that match those produced by people. We demonstrate that a perceptual relevance classifier is attainable and identify a good methodology to achieve this. The paper compares MLP, SVM, Bayes and regression classifiers for classifying shapes. MLPs perform best with an overall accuracy of 96.4%.

Original language	Undefined/Unknown
Pages	138-145
DOIs	https://doi.org/10.1145/1282280.1282306
Publication status	Published - 2007

Access to Document

10.1145/1282280.1282306

http://doi.acm.org/10.1145/1282280.1282306

Cite this

@conference{ab3727dbac134bb78eb7e933b5fdd401,

title = "Inducing a perceptual relevance shape classifier",

abstract = "In this paper, we develop a system to classify the outputs of image segmentation algorithms as perceptually relevant or perceptually irrelevant with respect to human perception. The work is aimed at figurative images. We previously investigated human visual perception of trademark images and established a body of ground truth data in the form of trademark images and their respective human segmentations. The work indicated that there is a core set of segmentations for each image that people perceive. Here we use this core set of segmentations to train a classifier to classify closed shapes output from an image segmentation algorithm so that the method returns the image segments that match those produced by people. We demonstrate that a perceptual relevance classifier is attainable and identify a good methodology to achieve this. The paper compares MLP, SVM, Bayes and regression classifiers for classifying shapes. MLPs perform best with an overall accuracy of 96.4%.",

author = "Hodge, {Victoria J.} and Eakins, {John P.} and James Austin",

year = "2007",

doi = "10.1145/1282280.1282306",

language = "Undefined/Unknown",

pages = "138--145",

}

TY - CONF

T1 - Inducing a perceptual relevance shape classifier

AU - Hodge, Victoria J.

AU - Eakins, John P.

AU - Austin, James

PY - 2007

Y1 - 2007

N2 - In this paper, we develop a system to classify the outputs of image segmentation algorithms as perceptually relevant or perceptually irrelevant with respect to human perception. The work is aimed at figurative images. We previously investigated human visual perception of trademark images and established a body of ground truth data in the form of trademark images and their respective human segmentations. The work indicated that there is a core set of segmentations for each image that people perceive. Here we use this core set of segmentations to train a classifier to classify closed shapes output from an image segmentation algorithm so that the method returns the image segments that match those produced by people. We demonstrate that a perceptual relevance classifier is attainable and identify a good methodology to achieve this. The paper compares MLP, SVM, Bayes and regression classifiers for classifying shapes. MLPs perform best with an overall accuracy of 96.4%.

AB - In this paper, we develop a system to classify the outputs of image segmentation algorithms as perceptually relevant or perceptually irrelevant with respect to human perception. The work is aimed at figurative images. We previously investigated human visual perception of trademark images and established a body of ground truth data in the form of trademark images and their respective human segmentations. The work indicated that there is a core set of segmentations for each image that people perceive. Here we use this core set of segmentations to train a classifier to classify closed shapes output from an image segmentation algorithm so that the method returns the image segments that match those produced by people. We demonstrate that a perceptual relevance classifier is attainable and identify a good methodology to achieve this. The paper compares MLP, SVM, Bayes and regression classifiers for classifying shapes. MLPs perform best with an overall accuracy of 96.4%.

U2 - 10.1145/1282280.1282306

DO - 10.1145/1282280.1282306

M3 - Paper

SP - 138

EP - 145

ER -