Research map / 3D vision / Depth and stereo
Refine and Distill: Exploiting Cycle-Inconsistency and Knowledge Distillation for Unsupervised Monocular Depth Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a backward network to form the cycle (ablation) | Half-cycle depth model · KITTI Eigen test split | Abs Rel: 0.1487 -> 0.1451 | Helped |
| Add inconsistency-aware refinement (ablation) | Cycle depth model · KITTI Eigen test split | Abs Rel: 0.1451 -> 0.1234 | Helped |
| Distill teacher disparities into the student (ablation) | Cycle depth model · KITTI Eigen test split, proposed reconstruction loss | Abs Rel: 0.1487 -> 0.1424 | Helped |
| Distill teacher features into the student (ablation) | Cycle depth model · KITTI Eigen test split, proposed reconstruction loss | Abs Rel: 0.1487 -> 0.1474 | Helped |
| Replace disparity distillation with feature distillation (ablation) | Student depth model · KITTI Eigen test split, proposed reconstruction loss | Abs Rel: 0.1424 -> 0.1474 | Hurt |
| Use the proposed two-network cycle instead of a single-network cycle (ablation) | Student depth model · KITTI Eigen test split | Abs Rel: 0.1503 -> 0.1438 | Helped |
| Use the proposed two-network cycle instead of a single-network cycle (ablation) | Teacher depth model · KITTI Eigen test split | Abs Rel: 0.1478 -> 0.1234 | Helped |
| Add a backward network to form the cycle (ablation) | Half-cycle depth model · Cityscapes test split | Abs Rel: 0.4676 -> 0.4523 | Helped |
| Add inconsistency-aware refinement (ablation) | Cycle depth model · Cityscapes test split | Abs Rel: 0.4523 -> 0.3988 | Helped |
| Use the proposed reconstruction loss instead of multiscale reconstruction loss (ablation) | Teacher depth model · KITTI Eigen test split, disparity distillation | Abs Rel: 0.1234 -> 0.0983 | Helped |
| Use the proposed reconstruction loss instead of multiscale reconstruction loss (ablation) | Teacher depth model · Cityscapes test split, disparity distillation | Abs Rel: 0.3988 -> 0.3846 | Helped |
Closest papers by meaning
- Learning Monocular Depth Estimation Infusing Traditional Stereo Knowledge (CVPR 2019 · similarity 0.90)
- Distilling Monocular Foundation Model for Fine-grained Depth Completion (CVPR 2025 · similarity 0.90)
- Self-Distilled Depth Refinement with Noisy Poisson Fusion (NeurIPS 2024 · similarity 0.88)
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · similarity 0.87)
- Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data (CVPR 2024 · similarity 0.87)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · similarity 0.87)
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.