Research map / 3D vision / Depth and stereo
Dense feature matching for correspondence
Learns dense or two-view correspondences with kernelized matching, cycle consistency and confidence estimation.
80 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Depth and stereo, keyword cluster “motion · flow · localization”
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.
23 papers in 2023–24 and 14 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 7.5% to 3.4%.
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 · Learning Dense Correspondence via 3D-Guided Cycle Consistency (CVPR 2016 · 362 citations)
- 2017 · Deep Multi-Scale Convolutional Neural Network for Dynamic Scene Deblurring (CVPR 2017 · 2,358 citations)
- 2018 · CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes (CVPR 2018 · 1,709 citations)
- 2019 · PlaneRCNN: 3D Plane Detection and Reconstruction From a Single Image (CVPR 2019 · 251 citations)
- 2020 · SuperGlue: Learning Feature Matching With Graph Neural Networks (CVPR 2020 · 2,845 citations)
- 2021 · Learning Accurate Dense Correspondences and When To Trust Them (CVPR 2021 · 127 citations)
- 2022 · Trusted Multi-View Deep Learning with Opinion Aggregation (AAAI 2022 · 79 citations)
- 2023 · DKM: Dense Kernelized Feature Matching for Geometry Estimation (CVPR 2023 · 172 citations)
- 2024 · View-decoupled Transformer for Person Re-identification under Aerial-ground Camera Network (CVPR 2024 · 46 citations)
- 2025 · SACB-Net: Spatial-awareness Convolutions for Medical Image Registration (CVPR 2025 · 13 citations)
- 2026 · Collaborative Feature Matching with Progressive Correspondence Learning (AAAI 2026 · 1 citation)
Earliest papers here
- Learning Dense Correspondence via 3D-Guided Cycle Consistency (CVPR 2016 · 362 citations)
- Pairwise Decomposition of Image Sequences for Active Multi-View Recognition (CVPR 2016 · 247 citations)
- Combining Fully Convolutional and Recurrent Neural Networks for 3D Biomedical Image Segmentation (NeurIPS 2016 · 220 citations)
What has been tried
73 papers here report 776 measured changes (a component added, removed or swapped, compared with a baseline): helped 76%, no clear effect 6%, hurt 15%, 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 |
|---|---|---|---|---|
| Remove hierarchical subtractive separation (Eq. 4) | CARGO and AG-ReID · performance (Rank1, mAP, and mINP) | Hurt | View-decoupled Transformer for Person Re-identification under Aerial-ground Camera Network (CVPR 2024 · 46 citations) | |
| Add BCMA to the iterative pruning strategy | YFCC100M unknown · mAP5° (%) | 51.80 -> 59.98 | Helped | BCLNet: Bilateral Consensus Learning for Two-View Correspondence Pruning (AAAI 2024 · 20 citations) |
| Train DeDoDe-B with a C4 steerer and rotation augmentation | MegaDepth-1500 · AUC @ 5° | 49 -> 51 | Helped | Steerers: A Framework for Rotation Equivariant Keypoint Descriptors (CVPR 2024 · 17 citations) |
| add VSFusion and ContextFormer to the two-iteration baseline | YFCC100M unknown scenes · mAP5° | 43.25 -> 62.18 | Helped | VSFormer: Visual-Spatial Fusion Transformer for Correspondence Pruning (AAAI 2024 · 14 citations) |
| add SACB at scales 5–2 | IXI · Dice | 0.7643 -> 0.7683 | Helped | SACB-Net: Spatial-awareness Convolutions for Medical Image Registration (CVPR 2025 · 13 citations) |
| Add ParamNet with non-linear equalization and random optical parameter dropout | D7000 sRGB→RAW · PSNR | 34.77 -> 36.21 | Helped | ParamISP: Learned Forward and Inverse ISPs using Camera Parameters (CVPR 2024 · 12 citations) |
| Add dual-branch SDNet instead of using single-branch decoding | TEyeD-subset A · 3D gaze error (°) | 1.18 -> 0.54 | Helped | De^2Gaze: Deformable and Decoupled Representation Learning for 3D Gaze Estimation (CVPR 2025 · 10 citations) |
| Add masked decomposition | YFCC100M · AUC@10° with weighted eight-point algorithm | 51.10 -> 52.67 | Helped | DeMatch: Deep Decomposition of Motion Field for Two-View Correspondence Learning (CVPR 2024 · 8 citations) |
| Add mutual supervision without stabilization or pre-training gaze constraint | Gaze360→ETH · Mono angular error | 18.73 -> 57.73 | Hurt | UVAGaze: Unsupervised 1-to-2 Views Adaptation for Gaze Estimation (AAAI 2024 · 7 citations) |
| Add PSV construction rather than process concatenated input images without a PSV | Sparse DTU, 9 views · PSNR | 17.30 -> 26.33 | Helped | Global Latent Neural Rendering (CVPR 2024 · 6 citations) |
| add the LCC block | YFCC100M · AUC@10° | 49.61 -> 54.82 | Helped | DeMo: Deep Motion Field Consensus with Learnable Kernels for Two-view Correspondence Learning (AAAI 2025 · 5 citations) |
| Use the personal branch alone instead of fusing both branches | TF2023 · IoU | 0.44 -> 0.40 | Hurt | Fusing Personal and Environmental Cues for Identification and Segmentation of First-Person Camera Wearers in Third-Person Views (CVPR 2024 · 3 citations) |
What reviewers will expect
Datasets and baselines named in the 9 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- RoMa · 3 of 9 papers
- LoFTR · 2 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 37 papers; from OpenAlex and the papers' affiliations).
- Wuhan University · 4 papers
- Tongji University · 3 papers
- Hangzhou Dianzi University · 2 papers
- The Chinese University of Hong Kong · 2 papers
Use it
Download BibTeX of 40 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
- Feed-forward 3D reconstruction (133 papers)
- Camera model and calibration for SfM (99 papers)
- Learned optical and scene flow (108 papers)
- Scene coordinate regression for localization (112 papers)
Elsewhere in the topic
- Self-supervised monocular depth (116 papers)
- Neural light field and illumination imaging (107 papers)
- Cost-volume multi-view stereo (107 papers)
- Non-line-of-sight imaging (93 papers)
- Dense depth prediction and completion (85 papers)
- Foundation models for monocular geometry (81 papers)
- Depth from light field and defocus (60 papers)
- Learned stereo with specialized sensors (53 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).