University of Notre Dame
Browse
- No file added yet -

Visual Psychophysics for Computer Vision

Download (152.72 MB)
thesis
posted on 2021-04-18, 00:00 authored by Brandon RichardWebster

By providing substantial amounts of data and standardized evaluation protocols, datasets in computer vision have helped fuel advances across all areas of visual recognition. But even in light of breakthrough results on recent benchmarks, it is still fair to ask if our recognition algorithms are doing as well as we think they are. The vision sciences at large make use of a very different evaluation regime known as Visual Psychophysics to study visual perception. Psychophysics is the quantitative examination of the relationships between controlled stimuli and the behavioral responses they elicit in experimental test subjects. Instead of using summary statistics to gauge performance, psychophysics directs us to construct item-response curves made up of individual stimulus responses to find perceptual thresholds, thus allowing one to identify the point at which a subject can no longer reliably recognize the stimulus.

In this work I will demonstrate the efficacy of visual psychophysics for computer vision from four main perspectives: visual psychophysics for evaluation, visual psychophysics for explainability, visual psychophysics for improved generalizability, and finally it presents a real life application that could substantially benefit from the capabilities visual psychophysics has to offer. The arch from evaluation to application completes a life cycle of a computer vision algorithm, providing positive evidence to the community that visual psychophysics can aid any stage of development.

History

Date Modified

2021-06-07

Defense Date

2021-02-19

CIP Code

  • 40.0501

Research Director(s)

Walter J. Scheirer

Committee Members

Adam Czajka Scott McCloskey Patrick Flynn

Degree

  • Doctor of Philosophy

Degree Level

  • Doctoral Dissertation

Alternate Identifier

1251804418

Library Record

6024127

OCLC Number

1251804418

Program Name

  • Computer Science and Engineering

Usage metrics

    Dissertations

    Categories

    No categories selected

    Keywords

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC