Research map / 3D vision / Depth and stereo
Tame a Wild Camera: In-the-Wild Monocular Camera Calibration
Shengjie Zhu, Abhinav Kumar, Masa Hu, Xiaoming Liu
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 |
|---|---|---|---|
| Use incidence-field calibration instead of Perspective intrinsic regression | Monocular camera calibration · NuScenes, KITTI, Cityscapes, NYUv2, ARKitScenes, SUN3D, MVImgNet, Objectron, MegaDepth, Waymo, RGBD, ScanNet, MVS, Scenes11 | e_f: NuScenes 0.378 -> 0.102; KITTI 0.631 -> 0.111; Cityscapes 0.624 -> 0.108; NYUv2 0.261 -> 0.086; ARKitScenes 0.325 -> 0.1 | Improved |
| Use incidence-field calibration instead of Perspective | Calibration on GSV · GSV | mean FoV error (°): 3.07 -> 2.49 | Improved |
| Use monocular calibration rather than FaceCalib | Camera calibration · BIWIRGBD-ID | e_f: 0.133 -> 0.034 | Improved |
| Use the proposed method rather than the direct-regression baseline | Geometric image manipulation restoration · KITTI | mIOU: 0.686 -> 0.842 | Improved |
| Assume a simple camera model during calibration (ablation) | Proposed monocular calibration · Datasets whose images fit the simple-camera assumption | e_f | Mixed |
| Estimate intrinsic from depthmap and surface-normal consistency rather than the incidence field (ablation) | Monocular camera calibration · In-the-wild images | intrinsic estimation stability | Hurt |
| Exclude SUN3D from training (ablation) | Proposed calibration model · Waymo versus RGBD, Scenes11, MVS, and ScanNet | e_f | Mixed |
Closest papers by meaning
- DiffCalib: Reformulating Monocular Camera Calibration as Diffusion-Based Dense Incident Map Generation (AAAI 2025 · similarity 0.89)
- MonoPlace3D: Learning 3D-Aware Object Placement for 3D Monocular Detection (CVPR 2025 · similarity 0.88)
- UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · similarity 0.88)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.87)
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.86)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.86)
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.
