Object recognition in humans depends primarily on shape cues. We have developed a new approach to measuring the shape recognition performance of a vision system based on nearest neighbor view matching within the system's embedding space. Our performance benchmark, ShapeY, allows for precise control of task difficulty, by enforcing that view matching span a specified degree of 3D viewpoint change and/or appearance change. As a first test case we measured the performance of ResNet50 pre-trained on ImageNet. Matching error rates were high. For example, a 27 degree change in object pitch led ResNet50 to match the incorrect object 45% of the time. Appearance changes were also highly disruptive. Examination of false matches indicates that ResNet50's embedding space is severely "tangled". These findings suggest ShapeY can be a useful tool for charting the progress of artificial vision systems towards human-level shape recognition capabilities.
@article{arxiv.2111.08174,
title = {ShapeY: Measuring Shape Recognition Capacity Using Nearest Neighbor Matching},
author = {Jong Woo Nam and Amanda S. Rios and Bartlett W. Mel},
journal= {arXiv preprint arXiv:2111.08174},
year = {2021}
}
Comments
6 pages, 5 figures, Accepted to NeurIPS: ImageNet Past, Present, and Future