# A Convolutional Decoder for Point Clouds using Adaptive Instance Normalization

arXiv: https://arxiv.org/abs/1906.11478 · HTML: https://mlsanity.com/paper/1906.11478/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Jun 2019.

## Built on by (most cited first, 8 of 8)

- [3D Shape Generation with Grid-based Implicit Functions](https://mlsanity.com/paper/2107.10607/) (2021)
- [Octree Transformer: Autoregressive 3D Shape Generation on Hierarchically Structured Sequences](https://mlsanity.com/paper/2111.12480/) (2021)
- [Voxel-based Network for Shape Completion by Leveraging Edge Generation](https://mlsanity.com/paper/2108.09936/) (2021)
- [PvDeConv: Point-Voxel Deconvolution for Autoencoding CAD Construction in 3D](https://mlsanity.com/paper/2101.04493/) (2021)
- [SoftPool++: An Encoder-Decoder Network for Point Cloud Completion](https://mlsanity.com/paper/2205.03899/) (2022)
- [Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning](https://mlsanity.com/paper/2403.05117/) (2024)
- [High-Fidelity Point Cloud Completion with Low-Resolution Recovery and Noise-Aware Upsampling](https://mlsanity.com/paper/2112.11271/) (2021)
- [CD2: Fine-grained 3D Mesh Reconstruction With Twice Chamfer Distance](https://mlsanity.com/paper/2206.00447/) (2022)

## Abstract (arXiv)

Automatic synthesis of high quality 3D shapes is an ongoing and challenging area of research. While several data-driven methods have been proposed that make use of neural networks to generate 3D shapes, none of them reach the level of quality that deep learning synthesis approaches for images provide. In this work we present a method for a convolutional point cloud decoder/generator that makes use of recent advances in the domain of image synthesis. Namely, we use Adaptive Instance Normalization and offer an intuition on why it can improve training. Furthermore, we propose extensions to the minimization of the commonly used Chamfer distance for auto-encoding point clouds. In addition, we show that careful sampling is important both for the input geometry and in our point cloud generation process to improve results. The results are evaluated in an auto-encoding setup to offer both qualitative and quantitative analysis. The proposed decoder is validated by an extensive ablation study and is able to outperform current state of the art results in a number of experiments. We show the applicability of our method in the fields of point cloud upsampling, single view reconstruction, and shape synthesis.