PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

High-level feature extraction using SVM with walk-based graph kernel
Jean-Philippe Vert, Tomoko Matsui, Shin'ichi Satoh and Yuji Uchiyama
In: Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2009). (2009) IEEE Computer Society , Washington, DC, USA , pp. 1121-1124. ISBN 978-1-4244-2353-8

Abstract

We investigate a method using support vector machines (SVMs) with walk-based graph kernels for high-level feature extraction from images. In this method, each image is first segmented into a finite set of homogeneous segments and then represented as a segmentation graph where each vertex is a segment and edges connect adjacent segments. Given a set of features associated with each segment, we then obtain a positive definite kernel between images by comparing walks in the respective segmentation graphs, and image classification is carried out with an SVM based on this kernel. In a benchmark experiment on the MediaMill challenge problem, the mean average precision increased from 0.216 (baseline) to 0.341 when our method was utilized.

EPrint Type:Book Section
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Machine Vision
Theory & Algorithms
ID Code:4438
Deposited By:Jean-Philippe Vert
Deposited On:13 March 2009