[en] The presentation will draw on an integrated approach to the field of computer vision based on three pillars: data, methods, and evaluation. In this field, the majority of scientific papers propose new methods using benchmark datasets and performance metrics that are justified by their applications.
In the presentation, we will address issues specifically related to the evaluation of methods. The goal is to encourage new practices and alternatives based on a rigorous theoretical framework. These aspects will be illustrated using the tasks of change detection and building damage assessment.
Research Center/Unit :
VIULab Montefiore Institute - Montefiore Institute of Electrical Engineering and Computer Science - ULiège
Disciplines :
Electrical & electronics engineering
Author, co-author :
Van Droogenbroeck, Marc ; Université de Liège - ULiège > Département d'électricité, électronique et informatique (Institut Montefiore) > Télécommunications
Language :
English
Title :
Toward a Better Understanding of Computer Vision Algorithm Performance
