Tensorflow implementation of Spectral-GANs for High-Resolution 3D Point-cloud Generation (IROS 2020)
Point-clouds are a popular choice for vision and graphics tasks due to their accurate shape description and direct acquisition from range-scanners. This demands the ability to synthesize and reconstruct high-quality point-clouds. Current deep generative models for 3D data generally work on simplified representations (e.g., voxelized objects) and cannot deal with the inherent redundancy and irregularity in point-clouds. A few recent efforts on 3D point-cloud generation offer limited resolution and their complexity grows with the increase in output resolution. In this work, we develop a principled approach to synthesize 3D point-clouds using a spectral-domain Generative Adversarial Network (GAN). This is the Tensorflow code of our paper.
- Set the input path to the ground truth spherical harmonics in main.py
- Give the ground truth point cloud path in spatial_train.py as a single numpy array (num_clouds, num_points, 3)
- Train:
python main.py --mode=train
- Evaluate:
python main.py --mode=evaluate
@article{ramasinghe2020spectral, title={Spectral-GANs for High-Resolution 3D Point-cloud Generation}, author={Ramasinghe, Sameera and Khan, Salman and Barnes, Nick and Gould, Stephen}, journal={IEEE/RSJ Intenrational Conference on Robots and Systems (IROS)}, year={2020} }