Research map / 3D vision / Depth and stereo
Iris: Bringing Real-World Priors into Diffusion Model for Monocular Depth Estimation
Xinhao Cai, Gensheng Pei, Zeren Sun, Yazhou Yao, Fumin Shen, Wenguan Wang
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 |
|---|---|---|---|
| replace multi-step stochastic diffusion with single-step deterministic prediction (ablation) | SD-based depth model · KITTI | AbsRel: 8.7 -> 8.3 | Improved |
| add SGD in single stage with simultaneous synthetic and real supervision (ablation) | deterministic model · KITTI | AbsRel: 8.3 -> 7.9 | Improved |
| use two-stage priors-to-geometry schedule with SGD instead of single stage (ablation) | deterministic model · KITTI | AbsRel: 7.9 -> 7.4 | Improved |
| add two-stage schedule alone without real-prior distillation (ablation) | deterministic model · KITTI | AbsRel: 8.3 -> 8.2 | No clear effect |
| replace vanilla distillation with spectral-gated distillation (ablation) | two-stage model · KITTI | AbsRel: 7.5 -> 7.4 | Improved |
| add spectral-gated consistency (SGC) (ablation) | two-stage + SGD model · KITTI | AbsRel: 7.4 -> 7.2 | Improved |
| add spectral-gated consistency (SGC) (ablation) | two-stage + SGD model · ScanNet | AbsRel: 5.2 -> 5.0 | Improved |
| add auxiliary image reconstruction loss (gamma=1) (ablation) | Iris loss · KITTI | AbsRel: 7.5 -> 7.4 | Improved |
| set SGC weight beta to 1 instead of 0.1 (ablation) | Iris loss · KITTI | AbsRel: 7.2 -> 7.9 | Hurt |
| distill from Depth Anything V2 teacher with 100K real images, versus training on synthetic data only | Iris · ETH3D | AbsRel: 6.1 -> 5.5 | Improved |
| use Iris instead of Lotus-D on DIODE | diffusion depth models · DIODE | AbsRel: 22.8 -> 24.3 | Hurt |
Closest papers by meaning
- Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (CVPR 2024 · similarity 0.90)
- BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation (NeurIPS 2024 · similarity 0.90)
- DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching (AAAI 2025 · similarity 0.89)
- Jasmine: Harnessing Diffusion Prior for Self-supervised Depth Estimation (NeurIPS 2025 · similarity 0.88)
- DCDepth: Progressive Monocular Depth Estimation in Discrete Cosine Domain (NeurIPS 2024 · similarity 0.88)
- Metric Depth Estimation from Arbitrarily Degraded Low-Resolution Depth Prompts (NeurIPS 2026 · 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.
