Solv3R: 3D Looped Transformer as a Multi-view Solver
Paper · https://arxiv.org/abs/2611.02857
Classical structure-from-motion reconstruction proceeds iteratively, refining camera poses and scene geometry from a current estimate. This allows maps to easily expand without recomputing already computed views, or geometry estimates to improve as more observations are provided. Most geometric foundation models, however, predict entire geometry in a single forward pass or update a recurrent state as frames arrive sequentially, making this expansion and refinement non-trivial. We introduce Solv3R, a looped transformer that learns to continue solving multi-view geometry from its current state. Importantly, its update operator is independent of the iteration count, allowing previously refined and newly introduced views to be jointly updated regardless of how ``solved'' they are. We train the operator to preserve accurate reconstructions under continued iteration and correct inaccurate states, enabling refinement of fixed image collections and integration of new observations. Representations of solved views can also be cached once and reused as geometric context across iterations, enabling compact and efficient key-value caching. We analyze the solver dynamics and compare it with the state-of-the-art, demonstrating that a single learned solver supports offline reconstruction, incremental reconstruction, relocalization, and streaming without task-specific training or modules. The code and models will be made publicly available at https://ubc-vision.github.io/solv3r/.
First seen by Oriel on .
AI summary
Solv3R is a looped transformer for multi-view geometry that iteratively refines existing reconstructions and incorporates new observations without task-specific modules.
- Its update operator is independent of iteration count, allowing solved and newly introduced views to be jointly updated.
- Training teaches the solver to preserve accurate reconstructions during continued iteration and correct inaccurate intermediate states.
- Cached representations of solved views provide geometric context across iterations through compact key-value caching.
- Experiments compare Solv3R with state-of-the-art methods for offline, incremental, relocalization, and streaming reconstruction.
- The code and models will be made publicly available.],
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hugging Face Papers: 17, latest observation ; 0 over observed period since
Comments
Comment count reported by the source.
- Hugging Face Papers: 3, latest observation ; 0 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Papers | 2611.02857 | https://arxiv.org/abs/2611.02857 Read discussion |