Giles Foody (University of Nottingham, UK)

Giles M. Foody is Professor of Geographical Information Science at the University of Nottingham, UK. His main research interests focus on the interface between remote sensing, ecology, and informatics.

Giles is founding editor-in-chief of Remote Sensing Letters and co-editor-in-chief of the International Journal of Remote Sensing, as well as a member of the editorial board of several other journals such as Remote Sensing of Environment, Landscape Ecology and Ecological Informatics.

Topics of particular interest relate to image classification for land cover mapping and monitoring applications, addressing issues at scales ranging from the sub-pixel to global.

Observations on accuracy assessments of object-based image classifications

This presentation will focus on the assessment of the accuracy of object-based image classifications. It will begin with an overview of the need for accuracy assessment. This will be followed by a summary of good practices for accuracy assessment, with particular regard to the three main stages encountered in traditional design-based assessments of accuracy: response design, sampling design and accuracy assessment. Using an example from each stage some key issues in accuracy assessment of object-based classifications will be highlighted. The presentation will show that assessments of the accuracy of object-based classifications are often flawed but hopefully illustrate how they could be enhanced.