Research map / 3D vision / Depth and stereo
Incremental Non-Rigid Structure-from-Motion with Unknown Focal Length
Thomas Probst, Danda Pani Paudel, Ajad Chhatkuli, Luc Van Gool
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 estimated focal length f_est instead of ground-truth focal length | template-based NRSfM · KINECT Paper | reconstruction error (mm): 3.00 -> 2.83 | Improved |
| use estimated focal length f_est instead of ground-truth focal length | template-based NRSfM · Hulk | reconstruction error (mm): 5.73 -> 5.53 | Improved |
| use estimated focal length f_est instead of ground-truth focal length | template-based NRSfM · Flag | reconstruction error (mm): 4.74 -> 4.54 | Improved |
| use estimated focal length f_est instead of ground-truth focal length | template-less NRSfM · Newspaper | reconstruction error (mm): 7.79 -> 9.27 | Hurt |
| add gradient-based refinement to the template-based focal estimate (ablation) | template-based calibration · Hulk | focal length error (%): 13.61 -> 8.85 | Improved |
| add gradient-based refinement to the template-based focal estimate (ablation) | template-based calibration · KINECT Paper | focal length error (%): 11.74 -> 14.45 | Hurt |
| add gradient-based refinement to the template-based focal estimate (ablation) | template-based calibration · Flag | reconstruction error (mm): 4.54 -> 4.34 | Improved |
| add gradient-based refinement from noisy initial K (+-20% noise) | calibration refinement · Hulk (simulated initial K) | focal length error (%): -2.36 | Improved |
| add gradient-based refinement from noisy initial K (+-20% noise) | calibration refinement · Flag (simulated initial K) | reconstruction error (mm): +0.08 | No clear effect |
| use incremental reconstruction (incr-tlmdh) instead of batch tlmdh | template-less NRSfM · KPaper | run time: 605.06s -> 176.16s | Improved |
| use incremental reconstruction (incr-tlmdh) instead of batch tlmdh | template-less NRSfM · KPaper | mean 3D error (mm): 5.41 -> 4.64 | Improved |
| use incremental reconstruction (incr-tlmdh) instead of batch tlmdh | template-less NRSfM · Hulk | mean 3D error (mm): 2.76 -> 2.99 | Hurt |
Closest papers by meaning
- Self-Calibrating Isometric Non-Rigid Structure-from-Motion (ECCV 2018 · similarity 0.92)
- Inextensible Non-Rigid Structure-from-Motion by Second-Order Cone Programming (TPAMI 2017 · similarity 0.92)
- Deep Non-Rigid Structure-from-Motion Revisited: Canonicalization and Sequence Modeling (AAAI 2025 · similarity 0.91)
- Robust Isometric Non-Rigid Structure-From-Motion (TPAMI 2021 · similarity 0.90)
- Non-Rigid Structure-from-Motion: Temporally-Smooth Procrustean Alignment and Spatially-Variant Deformation Modeling (CVPR 2024 · similarity 0.89)
- Structure-from-Motion with a Non-Parametric Camera Model (CVPR 2025 · similarity 0.88)
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.
