Position-Query-Based Autoencoders for View Decoupled Cross Point Cloud Reconstruction and a Self-Supervised Learning Framework

Xiangdong Zhang, Shaofeng Zhang, Junchi Yan

Keywords: Representation learning, Self-supervised learning, Point Cloud, Generative Methods

TL;DR: We introduce Point-PQAE, a novel generative framework for self-supervised learning of 3D point clouds that achieves cross-reconstruction between decoupled point clouds, enhancing both pre-training difficulty and performance.

Abstract: Point cloud learning, especially in a self-supervised way without manual labels, has received emerging attention in both vision and learning communities, with its potential utility in wide areas. Most existing generative approaches for point cloud self-supervised learning focus on recovering masked points from visible ones within a single view. Recognizing that a two-view pre-training paradigm inherently introduces greater diversity and variance, it could thus enable more challenging and informative pre-training. Inspired by this, we explore the potential of two-view learning in this domain. In this paper, we propose Point-PQAE, a cross-reconstruction generative paradigm that first generates two decoupled point clouds/views and then reconstructs one from the other. To achieve this goal, we develop a crop mechanism for point cloud view generation for the first time and further propose a novel positional encoding to represent the 3D relative position between the two decoupled views. The cross-reconstruction significantly increases the difficulty of pre-training compared to self-reconstruction, which enables our method to achieve new state-of-the-art results and surpasses previous single-modal self-reconstruction methods in 3D self-supervised learning by a margin. Specifically, it outperforms self-reconstruction baseline (Point-MAE) 6.5\%, 7.0\%, 6.7\% in three variants of ScanObjectNN with Mlp-Linear evaluation protocol. Source code will be released.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 4116