Kosmo-GFM: A Production-Ready Geometry Foundation Model
cs.CV
Submitted: 2026-11-12
Updated: 2026-11-12
Code: https://github.com/dazinovic/neural-rgbd-surface-reconstruction
Project page: https://kosmoresearch.github.io/Kosmo-GFM
Terminology
Sources
- L4P: Towards Unified Low-Level 4D Vision Perception
- Unified Panoramic Geometry Estimation via Multi-View Foundation Models
- D'ej\`a View: Looping Transformers for Multi-View 3D Reconstruction
- Virtual KITTI 2
- MUSt3R: Multi-view Network for Stereo 3D Reconstruction
- Any3R: Feedforward 3D Reconstruction from Any Camera Type
- Geometric Context Transformer for Streaming 3D Reconstruction
- RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation
- TTT3R: 3D Reconstruction as Test-Time Training
- LongStream: Long-Sequence Streaming Autoregressive Visual Geometry
- Glob3R: Global Structure-from-Motion with 3D Foundation Models
- SAIL-Recon: Large SfM by Augmenting Scene Regression with Localization
- VGGT-Long: Chunk it, Loop it, Align it -- Pushing VGGT's Limits on Kilometer-scale Long RGB Sequences
- Fisheye3R: Adapting Unified 3D Feed-Forward Foundation Models to Fisheye Lenses
- MASt3R-SfM: a Fully-Integrated Solution for Unconstrained Structure-from-Motion
- RoMa: Robust Dense Feature Matching
- DaD: Distilled Reinforcement Learning for Diverse Keypoint Detection
- RoMa v2: Harder Better Faster Denser Feature Matching
- VGG-T$^3$: Offline Feed-Forward 3D Reconstruction at Scale
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs