Research map / 3D vision / Depth and stereo
Leveraging Near-Field Lighting for Monocular Depth Estimation from Endoscopy Videos
Akshay Paruchuri, Samuel Ehrenstein, Shuxian Wang, Inbar Fried, Stephen Pizer, Marc Niethammer, Roni Sengupta
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 supervised PPS loss to the backbone (ablation) | Ours-Teacher · C3VD test | RMSE: 2.79 -> 2.21 | Improved |
| add PPSNet depth refinement module (ablation) | Ours-Teacher · C3VD test | RMSE: 2.79 -> 2.60 | Improved |
| add PPSNet on top of supervised PPS loss (ablation) | Ours-Teacher · C3VD test | AbsRel: 0.0518 -> 0.0529 | No clear effect |
| modulate with both RGB and PPS features instead of one (ablation) | PPSNet · C3VD test | RMSE: 2.29 -> 2.15 | Improved |
| add self-supervised PPS correlation loss on real clinical data (ablation) | Ours-Student · C3VD test | RMSE: 2.28 -> 2.06 | Improved |
| use the PPS-trained Teacher instead of a plain DINOv2 teacher (ablation) | Ours-Student · C3VD test | RMSE: 2.31 -> 2.06 | Improved |
| replace LightDepth rendering losses with PPS losses (ablation) | Ours-Student · C3VD test | RMSE: 2.29 -> 2.06 | Improved |
| teacher-student transfer to unlabeled clinical data | Ours-Teacher · C3VD test | RMSE: 2.15 -> 2.06 | Improved |
| DINOv2 encoder instead of DPT-Hybrid in the backbone (ablation) | Ours-Backbone · C3VD test | RMSE: 3.31 -> 2.79 | Improved |
Closest papers by meaning
- Depth Any Endoscopy: Towards Self-Supervised Generalizable Depth Estimation in Monocular Endoscopy (CVPR 2026 · similarity 0.92)
- DeLightMono: Enhancing Self-Supervised Monocular Depth Estimation in Endoscopy by Decoupling Uneven Illumination (AAAI 2026 · similarity 0.91)
- Unsupervised Photometric-Consistent Depth Estimation from Endoscopic Monocular Video (AAAI 2025 · similarity 0.90)
- LightDepth: Single-View Depth Self-Supervision from Illumination Decline (ICCV 2023 · similarity 0.88)
- NFL-BA: Near-Field Light Bundle Adjustment for SLAM in Dynamic Lighting (NeurIPS 2025 · similarity 0.85)
- Multi-view Masked Contrastive Representation Learning for Endoscopic Video Analysis (NeurIPS 2024 · 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.
