← Back to all news
arXiv Computer Vision By Gyeongjin Kang, Freeman Cheng, Botao Ye, Weijie Lyu, Ming-Hsuan Yang, Eunbyung Park

Rethinking Asymmetry in Self-Supervised 3D Geometry Learning

Read the original on arXiv Computer Vision →

The paper proposes a new self‑supervised approach for learning 3D geometry from unposed images by shifting the asymmetry from the input space to the 3D representation space. By excluding held‑out views only from the reconstructed scene while using a shared backbone for all views, the method improves depth and pose estimation. It further leverages view density differences and aligns features with a frozen vision foundation model to provide additional geometric training signals without requiring annotations.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

Related stories

arXiv AI
Oct 1

Emergent Multi-View Geometry Through Self-Distillation

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.

By David Nordstr\"om, Thibaut Loiseau, Vincent Lepetit, Michael Felsberg, Guillaume Bourmaud, Fredrik Kahl
computer-visionefficiency
More like this →
arXiv Computer Vision
Aug 28

DINOcular: Self-Supervised Visuospatial Representations

DINOcular is a self‑supervised framework that learns joint visuospatial representations from RGB‑D observations. It fuses depth‑derived geometric priors with a visual backbone using inter‑patch and intra‑patch fusion, allowing the model to encode both appearance and spatial structure efficiently. The resulting representation improves 3D awareness on multiple geometry benchmarks while staying competitive on standard RGB‑D semantic segmentation tasks.

By Farkhat Almukhamedov, Sami Azirar, Hermann Blum
computer-visionroboticsbenchmarks
More like this →
arXiv Computer Vision
Sep 3

Self-Geometry: GT-Free and Plug-and-Play Test-Time Adaptation for Geometrically Consistent 3D Vision Foundation Models

The paper introduces Self-Geometry, a plug‑and‑play test‑time adaptation framework that enforces explicit multi‑view geometric constraints on Vision Foundation Models (VFMs) using 2D pixel correspondences as pseudo ground truth. It combines Geometric Disentanglement Optimization—mixing Multi‑View and Epipolar Consistency losses with Gradient Disentanglement—to avoid gradient conflicts, a Frame Angular‑Neighbor sampler based on SO(3) geodesic distances to select informative views, and a Lightweight TTA module that adapts VFMs via LoRA. Experiments on six VFMs and four benchmarks (7Scenes, ETH3D, ScanNet++, HiRoom) show consistent improvements in pose and geometry estimation.

By Seokhyun Youn, Dahyeon Kye, Sung-Ho Bae, Jihyong Oh
fine-tuningbenchmarks
More like this →