Our world is full of identical objects (\emphe.g., cans of coke, cars of same model). These duplicates, when seen together, provide additional and strong cues for us to effectively reason about 3D. Inspired by this observation, we introduce Structure from Duplicates (SfD), a novel inverse graphics framework that reconstructs geometry, material, and illumination from a single image containing multiple identical objects. SfD begins by identifying multiple instances of an object within an image, and then jointly estimates the 6DoF pose for all instances.An inverse graphics pipeline is subsequently employed to jointly reason about the shape, material of the object, and the environment light, while adhering to the shared geometry and material constraint across instances. Our primary contributions involve utilizing object duplicates as a robust prior for single-image inverse graphics and proposing an in-plane rotation-robust Structure from Motion (SfM) formulation for joint 6-DoF object pose estimation. By leveraging multi-view cues from a single image, SfD generates more realistic and detailed 3D reconstructions, significantly outperforming existing single image reconstruction models and multi-view reconstruction approaches with a similar or greater number of observations.
@inproceedings{cheng2023structure,title={Structure from Duplicates: Neural Inverse Graphics from a Pile of Objects},author={Cheng, Tianhang and Ma, Wei-Chiu and Guan, Kaiyu and Torralba, Antonio and Wang, Shenlong},booktitle={Thirty-seventh Conference on Neural Information Processing Systems},year={2023},url={https://openreview.net/forum?id=7irm2VJARb},}
TGRS
(Under Review) Quantifying top-view crop residue cover from street-view imagery by proxy 3D field model
Tianhang Cheng, Sheng Wang, Kaiyu Guan, and 3 more authors
IEEE Transactions on Geoscience and Remote Sensing, 2023
2020
Designing conformal cloaks by manipulating structures directly in the physical space
Zhanyi Wang, Yichao Liu, Tianhang Cheng, and 2 more authors
The method to elaborately design the refractive index profile in the lower Riemann sheet of Zhukovski transformation plays an important role in the performance of this kind of conformal cloaks. However, for most proposed schemes, the mathematical calculations are complex. Here, we propose a more convenient method to design conformal cloaks by manipulating structures directly in the physical space. The designed cloak only needs symmetrical metal boundaries filled with normal dielectrics (refractive index ranges from 1 to 2) in the ‘circular branch cut’, which would be more feasible for future experimental implementation. Numerical simulations are performed by using the finite element method to validate our theoretical analysis.
@article{wang2020designing,title={Designing conformal cloaks by manipulating structures directly in the physical space},author={Wang, Zhanyi and Liu, Yichao and Cheng, Tianhang and Sun, Fei and He, Sailing},journal={Optics Express},volume={28},number={16},pages={23105--23113},year={2020},publisher={Optica Publishing Group},url={https://opg.optica.org/oe/fulltext.cfm?uri=oe-28-16-23105&id=433645},}
