Human expert fusion for image classification

: In image classification, merging the opinion of several human experts is very important for solving different tasks such as evaluation or training. Indeed, the reality is rarely known before scene imaging. The authors propose in this article different models in order to fuse the information given by two or more experts. The considered unit for classification, a small tile of the image, can contain one or more of the considered classes given by the experts. A second problem that has to be considered, is the degree of certainty of the expert s/he puts in each pixel of the tile. In order to solve these problems, the authors define five models in the context of Dempster-Shafer and Dezert-Smarandache theories and investigate the possible solutions with these models.

• Type: Journal
• Year: 2008
• Where: CORR
• Authors: Arnaud Martin, Christophe Osswald
• Keywords: CORR 2008, Education, Expert, Human Experts, Image Classification