TY - GEN
T1 - Using Shape to Categorize
T2 - 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021
AU - Stojanov, Stefan
AU - Thai, Anh
AU - Rehg, James M.
N1 - Funding Information:
We would like to thank Rohit Gajawada and Jiayuan Chen for their help with initial data collection. We also thank Zixuan Huang and Miao Liu for their helpful discussion of the paper draft. This work was supported by NSF award 1936970 and NIH award R01-MH114999.
Publisher Copyright:
© 2021 IEEE
PY - 2021
Y1 - 2021
N2 - It is widely accepted that reasoning about object shape is important for object recognition. However, the most powerful object recognition methods today do not explicitly make use of object shape during learning. In this work, motivated by recent developments in low-shot learning, findings in developmental psychology, and the increased use of synthetic data in computer vision research, we investigate how reasoning about 3D shape can be used to improve low-shot learning methods' generalization performance. We propose a new way to improve existing low-shot learning approaches by learning a discriminative embedding space using 3D object shape, and using this embedding by learning how to map images into it. Our new approach improves the performance of image-only low-shot learning approaches on multiple datasets. We also introduce Toys4K, a 3D object dataset with the largest number of object categories currently available, which supports low-shot learning.
AB - It is widely accepted that reasoning about object shape is important for object recognition. However, the most powerful object recognition methods today do not explicitly make use of object shape during learning. In this work, motivated by recent developments in low-shot learning, findings in developmental psychology, and the increased use of synthetic data in computer vision research, we investigate how reasoning about 3D shape can be used to improve low-shot learning methods' generalization performance. We propose a new way to improve existing low-shot learning approaches by learning a discriminative embedding space using 3D object shape, and using this embedding by learning how to map images into it. Our new approach improves the performance of image-only low-shot learning approaches on multiple datasets. We also introduce Toys4K, a 3D object dataset with the largest number of object categories currently available, which supports low-shot learning.
UR - http://www.scopus.com/inward/record.url?scp=85121413773&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85121413773&partnerID=8YFLogxK
U2 - 10.1109/CVPR46437.2021.00184
DO - 10.1109/CVPR46437.2021.00184
M3 - Conference contribution
AN - SCOPUS:85121413773
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 1798
EP - 1808
BT - Proceedings - 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021
PB - IEEE Computer Society
Y2 - 19 June 2021 through 25 June 2021
ER -