MoSE3: Learning World-Space SE(3) at Every Pixel
cs.CV
Submitted: 2026-10-02
Updated: 2026-10-02
Sources
- ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth
- St4RTrack: Simultaneous 4D Reconstruction and Tracking in the World
- DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos
- Syn4D: A Multiview Synthetic 4D Dataset
- Procedural Generation of Articulated Simulation-Ready Assets
- CoTracker3: Simpler and Better Point Tracking by Pseudo-Labelling Real Videos
- Any4D: Unified Feed-Forward Metric 4D Reconstruction
- MoSca: Dynamic Gaussian Fusion from Casual Videos via 4D Motion Scaffolds
- Infinite Mobility: Scalable High-Fidelity Synthesis of Articulated Objects via Procedural Generation
- Depth Anything 3: Recovering the Visual Space from Any Views
- Trace Anything: Representing Any Video in 4D via Trajectory Fields
- Track4World: Feedforward World-centric Dense 3D Tracking of All Pixels
- 4RC: 4D Reconstruction via Conditional Querying Anytime and Anywhere
- DELTA: Dense Efficient Long-range 3D Tracking for any video
- The 2017 DAVIS Challenge on Video Object Segmentation
- V-DPM: 4D Video Reconstruction with Dynamic Point Maps
- SceneTracker: Long-term Scene Flow Estimation Network
- VGGT: Visual Geometry Grounded Transformer
- Shape of Motion: 4D Reconstruction from a Single Video
- Continuous 3D Perception Model with Persistent State
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models