Testing conditions for viewpoint invariance in object recognition

William G. HAYWARD*, Michael J. TARR

*Corresponding author for this work

Research output: Journal PublicationsJournal Article (refereed)peer-review

108 Citations (Scopus)

Abstract

Based on the geon structural description approach, I. Biederman and P. C. Gerhardstein (1993) proposed 3 conditions under which object recognition is predicted to be viewpoint invariant. Two experiments are reported that satisfied all 3 criteria yet revealed performance that was clearly viewpoint dependent. Experiment 1 demonstrated that for both sequential matching and naming tasks, recognition of qualitatively distinct objects became progressively longer and less accurate as the viewpoint difference between study and test viewpoints increased. Experiment 2 demonstrated that for single-part objects, larger effects of viewpoint occurred when there was a change in the visible structure, indicating sensitivity to qualitative features in the image, not geon structural descriptions. These results suggest that the conditions proposed by I. Biederman and P. C. Gerhardstein are not generally applicable, the recognition of qualitatively distinct objects often relies on viewpoint-dependent mechanisms, and the molar features of view-based mechanisms appear to be image features rather than geons.

Original languageEnglish
Pages (from-to)1511-1521
Number of pages11
JournalJournal of Experimental Psychology: Human Perception and Performance
Volume23
Issue number5
DOIs
Publication statusPublished - Oct 1997
Externally publishedYes

Bibliographical note

This research was supported by a Yale university graduate fellowship and U.S. Office of Naval Research Contract N00014-93-1-0305.

Fingerprint

Dive into the research topics of 'Testing conditions for viewpoint invariance in object recognition'. Together they form a unique fingerprint.

Cite this