The paper identifies a theoretical incompatibility between 3D point cloud masked autoencoding and SE(3)-equivariant networks, showing that standard masking causes a topological collapse that forces models to hallucinate orientation. To address this, the authors introduce SIEVE, an SE(3)-equivariant self-supervised framework that compresses point cloud geometry into invariant scalar embeddings using a spherical projection mask that preserves global pose while obfuscating local semantics. Experiments demonstrate that SIEVE avoids the predicted collapse where constant and noise masks fail, and its embeddings outperform coordinate-based baselines and hand-crafted geometric descriptors on downstream point cloud tasks, especially in generalization to unseen shape categories.
We identify a theoretical incompatibility between 3D point cloud masked autoencoding and SE(3)-equivariant networks. We prove that standard masking triggers a topological collapse, effectively forcing...
arXiv:2606. 31570v1 Announce Type: cross Abstract: Masked autoencoding has emerged as a prominent paradigm for self-supervised learning on 3D point clouds, achieving competitive performance across downstream tasks.
arXiv:2603.25165v3 Announce Type: replace Abstract: Recent advances in self-supervised learning (SSL) for point clouds have substantially improved 3D scene understanding without human annotations. Ex...
arXiv:2608.24093v1 Announce Type: cross Abstract: Self-supervised representation learning for 4D point cloud videos is challenging because annotations are costly and reconstruction-based pretraining...
arXiv:2605. 17131v2 Announce Type: replace-cross Abstract: Point cloud stands as the most widely adopted format for representing 3D shapes and scenes due to its simplicity and geometric fidelity.
P-CORE introduces a self‑supervised surface consistency technique for point‑based neural representations, enabling robust adaptation to large deformations without needing ground‑truth deformed images. By generating random deformations and enforcing that the predicted surface after deformation matches the deformation applied to the original surface prediction, the method leverages attention‑based point representations with a learned interpolation kernel. Experiments on synthetic benchmarks and real‑world datasets show improved zero‑shot editing performance and reduced artifacts compared to existing point‑based approaches.
arXiv:2609.26039v1 Announce Type: cross Abstract: Point cloud generation has emerged as a crucial task for accurately capturing and reproducing the complexity of the physical world. However, existing...
arXiv:2610.03717v1 Announce Type: cross Abstract: This paper examines the role of Novel View Synthesis (NVS) in geometric representation learning. In principle, NVS should reason about 3D scene struc...
The paper introduces Poincar3, a self‑supervised method that learns multi‑view representations through self‑distillation rather than RGB reconstruction. By combining masked patch and image‑level distillation with a teacher that sees additional views, it trains from scratch without explicit 3D supervision. Poincar3 surpasses prior single‑ and multi‑view self‑supervised methods on tasks such as correspondence estimation, camera pose estimation, and 3D reconstruction, and its features encode camera motion more accurately thanks to a lightweight Poincaré adapter.
arXiv:2608. 05774v1 Announce Type: cross Abstract: Joint-embedding predictive architectures learn by predicting latent representations of missing observations, yet many masked JEPAs are evaluated primarily through the encoders they produce.
arXiv:2409. 07558v2 Announce Type: replace-cross Abstract: Rigid point cloud registration is a fundamental problem and highly relevant in robotics and autonomous driving.
The paper introduces a self‑supervised, geometry‑aware method that refines DINOv3 features with LoRA to better handle partial 3D shapes. By training on partial observations, the resulting descriptors outperform traditional and other vision‑foundation features in partial shape matching tasks. When incorporated into existing pipelines, the approach achieves state‑of‑the‑art results on both partial shape matching and left‑right prediction benchmarks.