Unsupervised Domain Adaptation for 3D Keypoint Prediction from a Single Depth Scan
Preprint
Abstract
In this paper, we introduce a novel unsupervised domain adaptation technique for the task of 3D keypoint prediction from a single depth scan/image. Our key idea is to utilize the fact that predictions from different views of the same or similar objects should be consistent with each other. Such view consistency provides effective regularization for keypoint prediction on unlabeled instances. In addition, we introduce a geometric alignment term to regularize predictions in the target domain. The resulting loss function can be effectively optimized via alternating minimization. We demonstrate the effectiveness of our approach on real datasets and present experimental results showing that our approach is superior to state-of-the-art general-purpose domain adaptation techniques.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks2018
- 3D Shape Segmentation with Projective Convolutional Networks2017
- Adversarial Discriminative Domain Adaptation2017
- Single Image 3D Interpreter Network2016
- View Synthesis by Appearance Flow2016
- Deep Sliding Shapes for Amodal 3D Object Detection in RGB-D Images2016
- Learning Deep Object Detectors from 3D Models2015
- 3D-Assisted Feature Synthesis for Novel Views of an Object2015
- 3D ShapeNets: A deep representation for volumetric shapes2015
- Viewpoints and keypoints2015