Research map / 3D vision / Human pose estimation
iMED: A Multi-Endoscope Dataset for Surgical 3D Perception
Sierra Bonilla, Fengyi Jiang, Chinedu Nwoye, Jingpei Lu, Kailey Reardon, Humphrey Chow, Francisco Vasconcelos, Sophia Bano, Adam Schmidt, Omid Mohareri
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 |
|---|---|---|---|
| add depth supervision (Depth Anything) (ablation) | NeRF · iMED rigid NVS, held-out endoscope | PSNR: 17.24 -> 16.29 | Hurt |
| add depth supervision (FoundationStereo) (ablation) | NeRF · iMED rigid NVS, held-out endoscope | PSNR: 17.24 -> 15.93 | Hurt |
| add depth supervision (FoundationStereo) (ablation) | 3DGS · iMED rigid NVS, held-out endoscope | PSNR / SSIM: PSNR 14.82 -> 15.33; SSIM 0.59 -> 0.63 | Improved |
| add depth supervision (Depth Anything) (ablation) | 3DGS · iMED rigid NVS, held-out endoscope | PSNR / SSIM: PSNR 14.82 -> 14.77; SSIM 0.59 -> 0.63 | Mixed |
| explicit dense 3D regularization (DynOmo) vs photometric-only D-3DGS | deformable NVS · iMED deformable NVS, held-out endoscope | PSNR: 14.50 -> 25.29 | Improved |
| swap depth prior from Depth Anything to FoundationStereo (ablation) | DynOmo · iMED online deformable NVS | PSNR: 24.16 -> 25.29 | Improved |
| swap depth prior from RAFT to FoundationStereo (ablation) | Hayoz · iMED online deformable NVS | PSNR: 15.66 -> 15.92 | No clear effect |
| dense foundation matcher vs sparse matcher for rotation accuracy | RoMa vs GIM+LightGlue · iMED pose estimation | rotation error (deg): 1.87 -> 1.79 | Improved |
| sparse learned matcher vs dense matcher for temporal consistency | GIM+LightGlue vs RoMa · iMED pose estimation | temporal drift (deg): 1.41 -> 0.41 | Improved |
| general-purpose metric depth model vs surgical-domain model | UniDepth-V2 vs Endo3R · iMED monocular depth (median scaled) | Abs Rel: 0.187 -> 0.081 | Improved |
| use predicted metric scale without median scaling (ablation) | UniDepth-V2 · iMED monocular depth | Abs Rel: 0.081 -> 11.573 | Hurt |
Closest papers by meaning
- EndoBench: A Comprehensive Evaluation of Multi-Modal Large Language Models for Endoscopy Analysis (NeurIPS 2025 · similarity 0.86)
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.86)
- Redefining the Laparoscopic Spatial Sense: AI-Based Intra- and Postoperative Measurement from Stereoimages (AAAI 2024 · similarity 0.85)
- MM-OR: A Large Multimodal Operating Room Dataset for Semantic Understanding of High-Intensity Surgical Environments (CVPR 2025 · similarity 0.85)
- Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering (CVPR 2024 · similarity 0.85)
- SurgVista: Long-Horizon Surgical World Modeling with Plausible Instrument-Tissue Dynamics (NeurIPS 2026 · similarity 0.84)
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.
