Download (3.8 MB)
Computer vision systems are traditionally tested in the object detection paradigm. In these experiments, a vision system is asked whether or not a specific object--for example an animal--occurs in a given image. A system that often answers correctly is said to be very accurate. In this talk, we will discuss some ambiguity that exists in this measure of accuracy. We will also propose a new measure of object-detection accuracy that addresses some of this ambiguity, and apply this measure to the hierarchical "standard model" of visual cortex.
Will Landecker obtained his B.A. in mathematics from Reed College, and is currently a PhD student in the PSU Computer Science program and a graduate research assistant at Los Alamos National Laboratory. He is conducting his research as a member of Melanie Mitchell's machine vision group. His research focuses on understanding the decisions of machine learning classifiers, particularly as they apply to computer vision systems. This work combines computer vision, theoretical machine learning, and data visualization. Other research interests include music informatics and computational neuroscience.
Computer vision, Image processing -- Digital techniques -- Evaluation, System theory, Visual cortex, Visualization -- Computer simulation
Landecker, Will; Thomure, Michael David; and Mitchell, Melanie, "Understanding Classification Decisions for Object Detection" (2010). Systems Science Friday Noon Seminar Series. 33.