# TearingNet: Point Cloud Autoencoder to Learn Topology-Friendly Representations

arXiv: https://arxiv.org/abs/2006.10187 · HTML: https://mlsanity.com/paper/2006.10187/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jun 2020.

## Built on by (most cited first, 23 of 23)

- [RGL-NET: A Recurrent Graph Learning framework for Progressive Part Assembly](https://mlsanity.com/paper/2107.12859/) (2021)
- [MPED: Quantifying Point Cloud Distortion based on Multiscale Potential Energy Discrepancy](https://mlsanity.com/paper/2103.02850/) (2021)
- [Flattening-Net: Deep Regular 2D Representation for 3D Point Cloud Analysis](https://mlsanity.com/paper/2212.08892/) (2022)
- [GRASP-Net: Geometric Residual Analysis and Synthesis for Point Cloud Compression](https://mlsanity.com/paper/2209.04401/) (2022)
- [Flatten Anything: Unsupervised Neural Surface Parameterization](https://mlsanity.com/paper/2405.14633/) (2024)
- [RefRec: Pseudo-labels Refinement via Shape Reconstruction for Unsupervised 3D Domain Adaptation](https://mlsanity.com/paper/2110.11036/) (2021)
- [Leveraging Single-View Images for Unsupervised 3D Point Cloud Completion](https://mlsanity.com/paper/2212.00564/) (2022)
- [PIVOT-Net: Heterogeneous Point-Voxel-Tree-based Framework for Point Cloud Compression](https://mlsanity.com/paper/2402.07243/) (2024)
- [Dynamic 3D Point Cloud Sequences as 2D Videos](https://mlsanity.com/paper/2403.01129/) (2024)
- [Arbitrary Point Cloud Upsampling with Spherical Mixture of Gaussians](https://mlsanity.com/paper/2208.05274/) (2022)
- [3D Shape Completion on Unseen Categories:A Weakly-supervised Approach](https://mlsanity.com/paper/2401.10578/) (2024)
- [Unsupervised 3D Point Cloud Completion via Multi-view Adversarial Learning](https://mlsanity.com/paper/2407.09786/) (2024)
- [Minimal Neural Atlas: Parameterizing Complex Surfaces with Minimal Charts and Distortion](https://mlsanity.com/paper/2207.14782/) (2022)
- [FlexPara: Flexible Neural Surface Parameterization](https://mlsanity.com/paper/2504.19210/) (2025)
- [PUMPS: Skeleton-Agnostic Point-based Universal Motion Pre-Training for Synthesis in Human Motion Tasks](https://mlsanity.com/paper/2507.20170/) (2025)
- [Learning Joint Surface Atlases](https://mlsanity.com/paper/2206.06273/) (2022)
- [ParaPoint: Learning Global Free-Boundary Surface Parameterization of 3D Point Clouds](https://mlsanity.com/paper/2403.10349/) (2024)
- [Generate Point Clouds with Multiscale Details from Graph-Represented Structures](https://mlsanity.com/paper/2112.06433/) (2021)
- [Variational Autoencoding of Dental Point Clouds](https://mlsanity.com/paper/2307.10895/) (2023)
- [WrappingNet: Mesh Autoencoder via Deep Sphere Deformation](https://mlsanity.com/paper/2308.15413/) (2023)
- [Point2Point : A Framework for Efficient Deep Learning on Hilbert sorted Point Clouds with applications in Spatio-Temporal Occupancy Prediction](https://mlsanity.com/paper/2306.16306/) (2023)
- [A 3D mesh convolution-based autoencoder for geometry compression](https://mlsanity.com/paper/2603.02125/) (2026)
- [SeamFlow: Structure-Aware Flow Matching on Edge Probabilities for Artist-Like UV Unwrapping](https://mlsanity.com/paper/2609.04751/) (2026)

## Abstract (arXiv)

Topology matters. Despite the recent success of point cloud processing with geometric deep learning, it remains arduous to capture the complex topologies of point cloud data with a learning model. Given a point cloud dataset containing objects with various genera, or scenes with multiple objects, we propose an autoencoder, TearingNet, which tackles the challenging task of representing the point clouds using a fixed-length descriptor. Unlike existing works directly deforming predefined primitives of genus zero (e.g., a 2D square patch) to an object-level point cloud, our TearingNet is characterized by a proposed Tearing network module and a Folding network module interacting with each other iteratively. Particularly, the Tearing network module learns the point cloud topology explicitly. By breaking the edges of a primitive graph, it tears the graph into patches or with holes to emulate the topology of a target point cloud, leading to faithful reconstructions. Experimentation shows the superiority of our proposal in terms of reconstructing point clouds as well as generating more topology-friendly representations than benchmarks.