TY - CONF
T1 - ADAPTIVE WAVELET TRANSFORMER NETWORK FOR 3D SHAPE REPRESENTATION LEARNING
AU - Huang, Hao
AU - Fang, Yi
N1 - Funding Information:
We would like to thank the reviewers and the authors of GDA-Net (Xu et al., 2021b) for their comments and efforts towards improving our manuscript. The authors appreciate the generous support provided by Inception Institute of Artificial Intelligence (IIAI) in the form of NYUAD Global Ph.D. Student Fellowship. This work was also partially supported by the NYUAD Center for Artificial Intelligence and Robotics (CAIR), funded by Tamkeen under the NYUAD Research Institute Award CG010.
Publisher Copyright:
© 2022 ICLR 2022 - 10th International Conference on Learning Representationss. All rights reserved.
PY - 2022
Y1 - 2022
N2 - We present a novel method for 3D shape representation learning using multi-scale wavelet decomposition. Previous works often decompose 3D shapes into complementary components in spatial domain at a single scale. In this work, we study to decompose 3D shapes into sub-bands components in frequency domain at multiple scales, resulting in a hierarchical decomposition tree in a principled manner rooted in multi-resolution wavelet analysis. Specifically, we propose Adaptive Wavelet Transformer Network (AWT-Net) that firstly generates approximation or detail wavelet coefficients per point, classifying each point into high or low sub-bands components, using lifting scheme at multiple scales recursively and hierarchically. Then, AWT-Net exploits Transformer to enhance the original shape features by querying and fusing features from different but integrated sub-bands. The wavelet coefficients can be learned without direct supervision on coefficients, and AWT-Net is fully differentiable and can be learned in an end-to-end fashion. Extensive experiments demonstrate that AWT-Net achieves competitive performance on 3D shape classification and segmentation benchmarks.
AB - We present a novel method for 3D shape representation learning using multi-scale wavelet decomposition. Previous works often decompose 3D shapes into complementary components in spatial domain at a single scale. In this work, we study to decompose 3D shapes into sub-bands components in frequency domain at multiple scales, resulting in a hierarchical decomposition tree in a principled manner rooted in multi-resolution wavelet analysis. Specifically, we propose Adaptive Wavelet Transformer Network (AWT-Net) that firstly generates approximation or detail wavelet coefficients per point, classifying each point into high or low sub-bands components, using lifting scheme at multiple scales recursively and hierarchically. Then, AWT-Net exploits Transformer to enhance the original shape features by querying and fusing features from different but integrated sub-bands. The wavelet coefficients can be learned without direct supervision on coefficients, and AWT-Net is fully differentiable and can be learned in an end-to-end fashion. Extensive experiments demonstrate that AWT-Net achieves competitive performance on 3D shape classification and segmentation benchmarks.
UR - http://www.scopus.com/inward/record.url?scp=85150383943&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85150383943&partnerID=8YFLogxK
M3 - Paper
AN - SCOPUS:85150383943
T2 - 10th International Conference on Learning Representations, ICLR 2022
Y2 - 25 April 2022 through 29 April 2022
ER -