Research map / 3D vision / Depth and stereo
Self-supervised monocular depth
Learns depth from unlabeled images or video using geometric constraints, semantic guidance and planarity priors.
116 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Depth and stereo, keyword cluster “depth estimation · monocular · depth map”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Still moving The best reported result is still improving on 2 of 4 tracked benchmarks.
29 papers in 2023–24 and 45 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 9.5% to 11.0%.
Open questions the data points to
- KITTI Eigen split (Abs Rel): best went from 0.039 (up to 2024) to 0.0251 in 2025 — Vision-Language Embodiment for Monocular Depth Estimation (CVPR 2025 · 5 citations)
- KITTI Eigen split (RMSE): best went from 1.743 (up to 2024) to 1.654 in 2025 — Vision-Language Embodiment for Monocular Depth Estimation (CVPR 2025 · 5 citations)
- Papers report DepthFormer on KITTI Eigen split (Abs Rel) anywhere from 0.052 to 0.09: comparisons against it depend on whose number you take.
- Papers report Monodepth2 on nuScenes (Abs Rel) anywhere from 0.287 to 0.425: comparisons against it depend on whose number you take.
What looks settled
- ETH3D (δ1): best up to 2024 98.1, best in 2025–26 41.5 from 2 papers, at most 0.2% better, if at all.
No longer reported by papers here: ETH3D (AbsRel, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Dense Monocular Depth Estimation in Complex Dynamic Scenes (CVPR 2016 · 191 citations)
- 2017 · Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · 3,325 citations)
- 2018 · Unsupervised Learning of Depth and Ego-Motion From Monocular Video Using 3D Geometric Constraints (CVPR 2018 · 833 citations)
- 2019 · Towards Scene Understanding: Unsupervised Monocular Depth Estimation With Semantic-Aware Representation (CVPR 2019 · 275 citations)
- 2020 · 3D Packing for Self-Supervised Monocular Depth Estimation (CVPR 2020 · 749 citations)
- 2021 · Learning To Recover 3D Scene Shape From a Single Image (CVPR 2021 · 212 citations)
- 2022 · P3Depth: Monocular Depth Estimation With a Piecewise Planarity Prior (CVPR 2022 · 159 citations)
- 2023 · iDisc: Internal Discretization for Monocular Depth Estimation (CVPR 2023 · 114 citations)
- 2024 · Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (CVPR 2024 · 388 citations)
- 2025 · UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · 26 citations)
- 2026 · GemDepth: Geometry-Embedded Features for 3D-Consistent Video Depth (ICML 2026 · 0 citations)
Earliest papers here
- Dense Monocular Depth Estimation in Complex Dynamic Scenes (CVPR 2016 · 191 citations)
- Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · 3,325 citations)
- Unsupervised Learning of Depth and Ego-Motion From Monocular Video Using 3D Geometric Constraints (CVPR 2018 · 833 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
KITTI Eigen split · Abs Rel (lower is better)
10 papers, 2017–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 0.114 | Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · 3,325 citations) |
| 2021 | 0.098 | The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · 16 citations) |
| 2022 | 0.054 | Multi-View Depth Estimation by Fusing Single-View Depth Probability With Multi-View Geometry (CVPR 2022 · 64 citations) |
| 2024 | 0.039 | Adaptive Fusion of Single-View and Multi-View Depth for Autonomous Driving (CVPR 2024 · 36 citations) |
| 2025 | 0.0251 | Vision-Language Embodiment for Monocular Depth Estimation (CVPR 2025 · 5 citations) |
KITTI Eigen split · RMSE (lower is better)
6 papers, 2020–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 4.485 | D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry (CVPR 2020 · 431 citations) |
| 2022 | 2.842 | P3Depth: Monocular Depth Estimation With a Piecewise Planarity Prior (CVPR 2022 · 159 citations) |
| 2024 | 1.743 | Adaptive Fusion of Single-View and Multi-View Depth for Autonomous Driving (CVPR 2024 · 36 citations) |
| 2025 | 1.654 | Vision-Language Embodiment for Monocular Depth Estimation (CVPR 2025 · 5 citations) |
ETH3D · AbsRel (lower is better)
4 papers, 2021–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 15.9 | Learning To Recover 3D Scene Shape From a Single Image (CVPR 2021 · 212 citations) |
| 2024 | 4.7 | BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation (NeurIPS 2024 · 0 citations) |
ETH3D · δ1 (higher is better)
4 papers, 2024–2025 · flat since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2024 | 98.1 | BetterDepth: Plug-and-Play Diffusion Refiner for Zero-Shot Monocular Depth Estimation (NeurIPS 2024 · 0 citations) |
What has been tried
86 papers here report 884 measured changes (a component added, removed or swapped, compared with a baseline): helped 73%, no clear effect 5%, hurt 20%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| use multi-resolution rather than Gaussian training noise | NYUv2 ablation split · AbsRel (%) | 7.7 -> 5.8 | Helped | Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (CVPR 2024 · 388 citations) |
| add the camera module | Out-of-Domain · δ1 | 48.38±0.97 -> 57.06±1.48 | Helped | UniDepth: Universal Monocular Metric Depth Estimation (CVPR 2024 · 189 citations) |
| Add the SQL layer with runtime-generated coarse-grained queries | KITTI Eigen · AbsRel | 0.114 -> 0.091 | Helped | SQLdepth: Generalizable Self-Supervised Fine-Structured Monocular Depth Estimation (AAAI 2024 · 58 citations) |
| Add feature guidance | UnrealStereo4K · REL | 0.0450 -> 0.0423 | Helped | PatchFusion: An End-to-End Tile-Based Framework for High-Resolution Monocular Metric Depth Estimation (CVPR 2024 · 44 citations) |
| Fuse single-view and multi-view depth with the adaptive fusion module | DDAD · SqRel | 1.59 -> 1.55 | Helped | Adaptive Fusion of Single-View and Multi-View Depth for Autonomous Driving (CVPR 2024 · 36 citations) |
| replace the pinhole camera representation with SH camera rays | L.FoV and Pano · FA | 45.2 -> 53.5; 24.6 -> 58.6 | Helped | UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · 26 citations) |
| set text-VAE optimization probability p to 0% rather than 1% | NYU Depth V2 · Abs Rel | 0.088 -> 0.091 | Hurt | WorDepth: Variational Language Prior for Monocular Depth Estimation (CVPR 2024 · 24 citations) |
| add pitch-aware ERP conversion | Pano3D-GV2 · δ1 | 0.4911 -> 0.7251 | Helped | Depth Any Camera: Zero-Shot Metric Depth Estimation from Any Camera (CVPR 2025 · 23 citations) |
| Increase native output resolution from 384×384 to 1536×1536 | UnrealStereo4K · boundary F1 | 0.044 -> 0.311 | Helped | Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · 20 citations) |
| Add ImageNet pre-training to ERP encoder | M3D · Abs Rel | 0.1596 -> 0.1115; 0.1211 -> 0.1048; 0.1515 -> 0.1423 | Helped | Elite360D: Towards Efficient 360 Depth Estimation via Semantic- and Distance-Aware Bi-Projection Fusion (CVPR 2024 · 18 citations) |
| Exclude dynamic-class objects from reprojection loss | Cityscapes WIR · abs rel | 0.125 -> 0.119 | Helped | From-Ground-To-Objects: Coarse-to-Fine Self-supervised Monocular Depth Estimation of Dynamic Objects with Ground Contact Prior (CVPR 2024 · 17 citations) |
| Reduce ensemble size from 5 to 1 | Replica · F-score | 0.853 -> 0.850 | No clear effect | Multi-view Reconstruction via SfM-guided Monocular Depth Estimation (CVPR 2025 · 12 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| DepthFormer | KITTI Eigen split | Abs Rel | 0.052 – 0.09 | 2 papers, e.g. CVPR 2024 and CVPR 2023 |
| Monodepth2 | nuScenes | Abs Rel | 0.287 – 0.425 | 2 papers, e.g. AAAI 2024 and NeurIPS 2023 |
What reviewers will expect
Datasets and baselines named in the 22 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- ScanNet · 6 of 22 papers
- KITTI · 6 of 22 papers
- Hypersim · 5 of 22 papers
- NYUv2 · 5 of 22 papers
- Booster · 4 of 22 papers
- ETH3D · 3 of 22 papers
- ScanNetv2 · 2 of 22 papers
- 7-Scenes · 2 of 22 papers
Baselines
- Metric3Dv2 · 8 of 22 papers
- Depth Anything V2 · 7 of 22 papers
- UniDepth · 5 of 22 papers
- Depth Anything · 5 of 22 papers
- ZoeDepth · 5 of 22 papers
- Marigold · 4 of 22 papers
- DepthPro · 4 of 22 papers
- HRDepth · 3 of 22 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 74 papers; from OpenAlex and the papers' affiliations).
- Meta · 3 papers
- Zhejiang University · 3 papers
- ETH Zurich · 3 papers
- Southeast University · 2 papers
- Northwestern Polytechnical University · 2 papers
- Tsinghua University · 2 papers
- Korea Advanced Institute of Science and Technology · 2 papers
- Princeton University · 2 papers
- Huazhong University of Science and Technology · 2 papers
- NVIDIA · 2 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Dense depth prediction and completion (85 papers)
- Foundation models for monocular geometry (81 papers)
- Depth from light field and defocus (60 papers)
Elsewhere in the topic
- Feed-forward 3D reconstruction (133 papers)
- Scene coordinate regression for localization (112 papers)
- Learned optical and scene flow (108 papers)
- Neural light field and illumination imaging (107 papers)
- Cost-volume multi-view stereo (107 papers)
- Camera model and calibration for SfM (99 papers)
- Non-line-of-sight imaging (93 papers)
- Dense feature matching for correspondence (80 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).