Research map / 3D vision / Depth and stereo
GeoCalib: Learning Single-image Calibration with Geometric Optimization
Alexander Veicht, Paul-Edouard Sarlin, Philipp Lindenberger, Marc Pollefeys
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train the network end-to-end by supervising the LM optimization output (ablation) | GeoCalib · OpenPano test | Gravity median error (deg): 3.03 -> 2.65 | Improved |
| add learned uncertainties (confidences) to the perspective field (ablation) | GeoCalib · OpenPano test | Gravity median error (deg): 2.65 -> 2.24 | Improved |
| replace SegFormer backbone with SegNeXt (ablation) | GeoCalib · OpenPano test | Gravity median error (deg): 2.24 -> 1.88 | Improved |
| replace ParamNet regression head with second-order LM optimization (no end-to-end training) (ablation) | PerspectiveNet+ParamNet · OpenPano test | FoV median error (deg): 4.23 -> 4.58 in-domain; MegaDepth 11.23 -> 8.70 | Mixed |
| replace Adam/SGD optimization with LM optimization out of the box (ablation) | GeoCalib ablation · OpenPano test | Gravity median error (deg): 3.03 -> 3.03 | No clear effect |
| train on OpenPano (about 5x smaller) instead of 360 Cities (ablation) | PerspectiveNet+ParamNet · MegaDepth | Gravity median error (deg): OpenPano 4.99 -> 3.08; MegaDepth 4.06 -> 4.51 | Mixed |
| use heuristic or solver initialization instead of trivial initialization (ablation) | GeoCalib · OpenPano test | Gravity median error (deg): 1.88 -> 1.90 (heuristic), 2.03 (solver) | No clear effect |
| fix known FoV in the optimization (partial calibration) (ablation) | GeoCalib · Stanford2D3D | Gravity median error (deg): 1.12 -> 0.86 | Improved |
| fix known gravity in the optimization (partial calibration) (ablation) | GeoCalib · Stanford2D3D | FoV median error (deg): 3.21 -> 2.30 | Improved |
| jointly optimize intrinsics over multiple images from the same camera (ablation) | GeoCalib · TartanAir | AUC of FoV and gravity | Improved |
| train GeoCalib on distorted images rather than pinhole only (ablation) | GeoCalib · MegaDepth distortion | pixel distortion recall @5px: 72.01 -> 75.57 | Improved |
| add estimated gravity prior in RANSAC and refinement | SP+SG localization · InLoc DUC1 | pose recall @0.25m,10deg: 43.4 -> 44.9 | Improved |
Later papers that compared with this method
3 later accepted papers report GeoCalib as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- InFlux++: Real and Synthetic Data for Estimating Dynamic Camera Intrinsics (ECCV 2026 · better on 2 of 2 shared results)
- AnyCalib: On-Manifold Learning for Model-Agnostic Single-View Camera Calibration (ICCV 2025 · better on 5 of 5 shared results)
- Boost 3D Reconstruction using Diffusion-based Monocular Camera Calibration (ICCV 2025 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Deep Geometry-Aware Camera Self-Calibration from Video (ICCV 2023 · similarity 0.88)
- AnyCalib: On-Manifold Learning for Model-Agnostic Single-View Camera Calibration (ICCV 2025 · similarity 0.88)
- GeoDepth: From Point-to-Depth to Plane-to-Depth Modeling for Self-Supervised Monocular Depth Estimation (CVPR 2025 · similarity 0.86)
- GeoRouter: Dynamic Paradigm Routing for Worldwide Image Geolocalization (NeurIPS 2026 · similarity 0.86)
- DiffCalib: Reformulating Monocular Camera Calibration as Diffusion-Based Dense Incident Map Generation (AAAI 2025 · similarity 0.86)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
