Domain Decluttering: Simplifying Images to Mitigate Synthetic-Real Domain Shift and Improve Depth Estimation

@article{Zhao2020DomainDS,
  title={Domain Decluttering: Simplifying Images to Mitigate Synthetic-Real Domain Shift and Improve Depth Estimation},
  author={Yunhan Zhao and Shu Kong and Daeyun Shin and Charless C. Fowlkes},
  journal={2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2020},
  pages={3327-3337}
}
Leveraging synthetically rendered data offers great potential to improve monocular depth estimation and other geometric estimation tasks, but closing the synthetic-real domain gap is a non-trivial and important task. While much recent work has focused on unsupervised domain adaptation, we consider a more realistic scenario where a large amount of synthetic training data is supplemented by a small set of real images with ground-truth. In this setting, we find that existing domain translation… Expand
3 Citations
When Perspective Comes for Free: Improving Depth Prediction with Camera Pose Encoding
  • 1
S2R-DepthNet: Learning a Generalizable Depth-specific Structural Representation
  • Xiaotian Chen, Yuwang Wang, Xuejin Chen, Wenjun Zeng
  • Computer Science
  • ArXiv
  • 2021
  • PDF
When Autonomous Systems Meet Accuracy and Transferability through AI: A Survey
  • 1
  • PDF

References

SHOWING 1-10 OF 66 REFERENCES
T2Net: Synthetic-to-Realistic Translation for Solving Single-Image Depth Estimation Tasks
  • 70
  • Highly Influential
  • PDF
Geometry-Aware Symmetric Domain Adaptation for Monocular Depth Estimation
  • 51
  • PDF
CrDoCo: Pixel-Level Domain Transfer With Cross-Domain Consistency
  • 96
  • PDF
Digging Into Self-Supervised Monocular Depth Estimation
  • 349
  • PDF
CGIntrinsics: Better Intrinsic Image Decomposition through Physically-Based Rendering
  • 53
  • PDF
Learning to Adapt Structured Output Space for Semantic Segmentation
  • 439
  • PDF
Deeper Depth Prediction with Fully Convolutional Residual Networks
  • 920
  • PDF
Learning from Synthetic Humans
  • 427
  • PDF
VirtualWorlds as Proxy for Multi-object Tracking Analysis
  • 536
  • PDF
...
1
2
3
4
5
...