Research map / 3D vision / 3D generation
MBW: Multi-view Bootstrapping in the Wild
Mosam Dabhi, Chaoyang Wang, Tim Clifford, Laszlo Attila Jeni, Ian R. Fasel, Simon Lucey
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 MV-NRSfM instead of calibrated multi-view triangulation for geometric rejection and denoising | MBW on Human3.6M Directions-1, Subject 1 · Human3.6M Directions-1, Subject 1 | 2D PCK AUC: 69.72 -> 81.41 | Improved |
| Use MV-NRSfM instead of calibrated multi-view triangulation for geometric rejection and denoising | MBW on Human3.6M Directions-1, Subject 1 · Human3.6M Directions-1, Subject 1 | 3D error (mm): 1383 -> 39.3 | Improved |
| Use MBW with 2% input labels | 3D pose reconstruction on Human3.6M · Human3.6M | PA-MPJPE (mm): 20.15 vs 17.5 | Mixed |
| Use MBW with 2% labeled data for semi-supervised landmark prediction | Human3.6M · Human3.6M, wrist | PCKh: 49.6 -> 69.8 | Improved |
| Continue MBW self-training from iteration 1 through iteration 2 (ablation) | MBW on Human3.6M Directions-1, Subject 1 · Human3.6M Directions-1, Subject 1 | 3D error (mm): 33.3 -> 28.3 | Improved |
| Continue MBW self-training from iteration 2 through iteration 3 (ablation) | MBW on Human3.6M Directions-1, Subject 1 · Human3.6M Directions-1, Subject 1 | 3D error (mm): 28.3 -> 23.1 | Improved |
| Use iterative pseudo-labeling and retraining rather than stopping after the initial stage (ablation) | MBW on Human3.6M · Human3.6M | 2D landmark prediction error | Improved |
Closest papers by meaning
- Self-supervised Secondary Landmark Detection via 3D Representation Learning (IJCV 2023 · similarity 0.89)
- BigMaQ: A Big Macaque Motion and Animation Dataset Bridging Image and 3D Pose Representations (ICLR 2026 · similarity 0.89)
- BEAST3D: animal behavioral analysis and neural encoding from multi-view video via Gaussian splatting (NeurIPS 2026 · similarity 0.88)
- APT-36K: A Large-scale Benchmark for Animal Pose Estimation and Tracking (NeurIPS 2022 · similarity 0.87)
- Leader360V: A Large-scale, Real-world 360 Video Dataset for Multi-task Learning in Diverse Environment (NeurIPS 2025 · similarity 0.85)
- Reconstructing Animals and the Wild (CVPR 2025 · 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.
