Research map / 3D vision / Depth and stereo
Depth from light field and defocus
Estimates depth using light fields, defocus, event cues and interference instead of standard RGB.
60 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 Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
15 papers in 2023–24 and 27 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 4.9% to 6.6%.
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 · Robust Light Field Depth Estimation for Noisy Scene With Occlusion (CVPR 2016 · 103 citations)
- 2017 · Depth From Defocus in the Wild (CVPR 2017 · 50 citations)
- 2018 · Trust Your Model: Light Field Depth Estimation With Inline Occlusion Handling (CVPR 2018 · 60 citations)
- 2019 · Depth-Attentional Features for Single-Image Rain Removal (CVPR 2019 · 410 citations)
- 2020 · From Depth What Can You See? Depth Completion via Auxiliary Image Reconstruction (CVPR 2020 · 89 citations)
- 2021 · Attention-based Multi-Level Fusion Network for Light Field Depth Estimation (AAAI 2021 · 93 citations)
- 2022 · Occlusion-Aware Cost Constructor for Light Field Depth Estimation (CVPR 2022 · 107 citations)
- 2023 · BEVDepth: Acquisition of Reliable Depth for Multi-View 3D Object Detection (AAAI 2023 · 757 citations)
- 2024 · Robust Depth Enhancement via Polarization Prompt Fusion Tuning (CVPR 2024 · 17 citations)
- 2025 · Prompting Depth Anything for 4K Resolution Accurate Metric Depth Estimation (CVPR 2025 · 29 citations)
- 2026 · Depth Anything with Any Prior (ICLR 2026 · 0 citations)
Earliest papers here
- Robust Light Field Depth Estimation for Noisy Scene With Occlusion (CVPR 2016 · 103 citations)
- Depth From Defocus in the Wild (CVPR 2017 · 50 citations)
- Noise Robust Depth From Focus Using a Ring Difference Filter (CVPR 2017 · 36 citations)
What has been tried
42 papers here report 377 measured changes (a component added, removed or swapped, compared with a baseline): helped 76%, no clear effect 5%, hurt 16%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add low-cost LiDAR prompting | ARKitScenes · L1 | 0.0605 -> 0.0138 | Helped | Prompting Depth Anything for 4K Resolution Accurate Metric Depth Estimation (CVPR 2025 · 29 citations) |
| use PPFT with polarization guidance instead of RGB guidance | HAMMER All sensors · RMSE | 33.06 -> 28.63 | Helped | Robust Depth Enhancement via Polarization Prompt Fusion Tuning (CVPR 2024 · 17 citations) |
| replace the graph-based Radar structure extractor with an MLP | nuScenes, 0–70m · MAE | 1712.6 -> 1815.6 | Hurt | TacoDepth: Towards Efficient Radar-Camera Depth Estimation with One-stage Fusion (CVPR 2025 · 8 citations) |
| Add random depth augmentation | NYUv2 sparsity bias · RMSE | 0.498 -> 0.231 | Helped | Depth Prompting for Sensor-Agnostic Depth Estimation (CVPR 2024 · 8 citations) |
| retain full RCCB image resolution | Night, LiDAR ground-truth points · MAE | 2.04 -> 2.03 | No clear effect | Cross-spectral Gated-RGB Stereo Depth Estimation (CVPR 2024 · 7 citations) |
| Use patch size 11 × 11 instead of 21 × 21 | Synthetic testing set · RMSE (cm) | 5.281 -> 5.675 | Hurt | Blurry-Edges: Photon-Limited Depth Estimation from Defocused Boundaries (CVPR 2025 · 4 citations) |
| Add FGDE | Compressed-NYU · MAE | Helped | Decoupling Fine Detail and Global Geometry for Compressed Depth Map Super-Resolution (CVPR 2025 · 4 citations) | |
| Add SFE to LMM | iToF test set · MAEd (cm) | 6.24 -> 5.83 | Helped | iToF-flow-based High Frame Rate Depth Imaging (CVPR 2024 · 4 citations) |
| Add the proposed temporal representation alignment instead of unaligned E2VID reconstruction | interlaken_00_c (DSEC) · EPE | 43.02 -> 1.49 | Helped | Zero-Shot Event-Intensity Asymmetric Stereo via Visual Prompting from Image Domain (NeurIPS 2024 · 3 citations) |
| add Smoothness Loss with Highlight Removal (LS) | C3VD · AbsRel | 0.082 -> 0.076 | Helped | Unsupervised Photometric-Consistent Depth Estimation from Endoscopic Monocular Video (AAAI 2025 · 2 citations) |
| add photometric reprojection loss | UIEB_dark · PSNR | 16.33 -> 17.21 | Helped | Sea-ing in Low-light (CVPR 2025 · 2 citations) |
| add spatial and focal variational constraints | NYU Depth v2 · RMSE | 0.094 -> 0.075 | Helped | DualFocus: Depth from Focus with Spatio-Focal Dual Variational Constraints (NeurIPS 2025 · 2 citations) |
What reviewers will expect
Datasets and baselines named in the 13 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- ARKitScenes · 2 of 13 papers
- nuScenes · 2 of 13 papers
- ZJU-4DRadarCam · 2 of 13 papers
Baselines
- Depth Anything · 3 of 13 papers
- Depth Anything V2 · 3 of 13 papers
- Marigold · 3 of 13 papers
- DefocusNet · 2 of 13 papers
- ZoeDepth · 2 of 13 papers
- DepthPro · 2 of 13 papers
- Lin et al. · 2 of 13 papers
- RC-PDA · 2 of 13 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 42 papers; from OpenAlex and the papers' affiliations).
- Shanghai Jiao Tong University · 2 papers
- Nanjing University of Science and Technology · 2 papers
- Peking University · 2 papers
- National University of Singapore · 2 papers
Use it
Download BibTeX of 39 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
- Self-supervised monocular depth (116 papers)
- Dense depth prediction and completion (85 papers)
- Foundation models for monocular geometry (81 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).