Research map / 3D vision / 3D generation
Monocular 3D object reconstruction
Reconstructs objects and dynamic scenes from monocular input using synthetic data and pose-free online methods.
94 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in 3D generation, keyword cluster “motion · interaction · human”
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.
26 papers in 2023–24 and 39 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 5.7% to 4.4%.
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
- 2017 · Hand Keypoint Detection in Single Images Using Multiview Bootstrapping (CVPR 2017 · 1,182 citations)
- 2018 · Now You Shake Me: Towards Automatic 4D Cinema (CVPR 2018 · 17 citations)
- 2019 · Volumetric Capture of Humans With a Single RGBD Camera via Semi-Parametric Learning (CVPR 2019 · 41 citations)
- 2020 · ARShadowGAN: Shadow Generative Adversarial Network for Augmented Reality in Single Light Scenes (CVPR 2020 · 123 citations)
- 2021 · GeoSim: Realistic Video Simulation via Geometry-Aware Composition for Self-Driving (CVPR 2021 · 91 citations)
- 2022 · ABO: Dataset and Benchmarks for Real-World 3D Object Understanding (CVPR 2022 · 154 citations)
- 2023 · OmniObject3D: Large-Vocabulary 3D Object Dataset for Realistic Perception, Reconstruction and Generation (CVPR 2023 · 188 citations)
- 2024 · MVHumanNet: A Large-scale Dataset of Multi-view Daily Dressing Human Captures (CVPR 2024 · 24 citations)
- 2025 · Mono3DVLT: Monocular-Video-Based 3D Visual Language Tracking (CVPR 2025 · 9 citations)
- 2026 · 4RC: 4D Reconstruction via Conditional Querying Anytime and Anywhere (ICML 2026 · 0 citations)
Earliest papers here
- Hand Keypoint Detection in Single Images Using Multiview Bootstrapping (CVPR 2017 · 1,182 citations)
- DUST: Dual Union of Spatio-Temporal Subspaces for Monocular Multiple Object 3D Reconstruction (CVPR 2017 · 32 citations)
- Now You Shake Me: Towards Automatic 4D Cinema (CVPR 2018 · 17 citations)
What has been tried
67 papers here report 623 measured changes (a component added, removed or swapped, compared with a baseline): helped 78%, no clear effect 4%, hurt 14%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add TrajControl to the iterative inference scheme | PROX RGB-D · Skating | 0.056 -> 0.038 | Helped | RoHM: Robust Human Motion Reconstruction via Diffusion (CVPR 2024 · 20 citations) |
| Remove flow loss | DAVIS · EPE (vis) | 8.56 -> 10.91 | Hurt | DreamScene4D: Dynamic Multi-Object Scene Generation from Monocular Videos (NeurIPS 2024 · 15 citations) |
| Add Manhattan regularization to the background reconstruction | Replica room 1 background reconstruction · background geometry regularity | Helped | Total-Decom: Decomposed 3D Scene Reconstruction with Minimal Interaction (CVPR 2024 · 10 citations) | |
| Replace ResNet50 with Swin Transformer features | Mono3DVLT-V2X test · SR@0.9 | 49.13 -> 53.91 | Helped | Mono3DVLT: Monocular-Video-Based 3D Visual Language Tracking (CVPR 2025 · 9 citations) |
| replace active-sensor SDF volume rendering with passive-sensor SDF volume rendering | TownClean · MAE | -1.5 | Hurt | Dynamic LiDAR Re-simulation using Compositional Neural Fields (CVPR 2024 · 9 citations) |
| Use fewer chunks (more temporal information) | DiVa-360 object and interaction sequences · PSNR | Helped | DiVa-360: The Dynamic Visual Dataset for Immersive Neural Fields (CVPR 2024 · 8 citations) | |
| Add amodal mask supervision | HO3D · CD | 1.094 -> 0.448 | Helped | In-Hand 3D Object Reconstruction from a Monocular RGB Video (AAAI 2024 · 7 citations) |
| Add a Kalman filter motion update | IT3DEgo SVOE validation · precision@0.25m | 23.3 -> 23.7 | Helped | Instance Tracking in 3D Scenes from Egocentric Videos (CVPR 2024 · 7 citations) |
| add LPIPS perceptual loss | MVImgNet test · LPIPS | 0.46 -> 0.29 | Helped | NViST: In the Wild New View Synthesis from a Single Image with Transformers (CVPR 2024 · 7 citations) |
| add dynamics conditioning | ZJU-MoCap (S313) · LPIPS | .120 -> .085 | Helped | SurMo: Surface-based 4D Motion Modeling for Dynamic Human Rendering (CVPR 2024 · 6 citations) |
| add Gaussian splatting to a textured mesh | Novel view synthesis · PSNR | 19.71 -> 25.87 | Helped | DRAWER: Digital Reconstruction and Articulation With Environment Realism (CVPR 2025 · 5 citations) |
| Use Morton parameterization instead of y-axis order or random mapping | 500-frame ablation · Rendering quality | Helped | RePerformer: Immersive Human-centric Volumetric Videos from Playback to Photoreal Reperformance (CVPR 2025 · 5 citations) |
What reviewers will expect
Datasets and baselines named in the 18 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- StanfordORB · 4 of 18 papers
- Objaverse · 2 of 18 papers
- CO3Dv2 · 2 of 18 papers
- PointOdyssey · 2 of 18 papers
Baselines
- 3DGS · 2 of 18 papers
- 2DGS · 2 of 18 papers
- NVDiffRec · 2 of 18 papers
- NVDiffRecMC · 2 of 18 papers
- InvRender · 2 of 18 papers
- PhySG · 2 of 18 papers
- NeRD · 2 of 18 papers
- DUSt3R · 2 of 18 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 65 papers; from OpenAlex and the papers' affiliations).
- Meta · 6 papers
- University of Oxford · 3 papers
- Carnegie Mellon University · 3 papers
- Adobe · 3 papers
- Massachusetts Institute of Technology · 3 papers
- Apple · 2 papers
- Department of Computer Science · 2 papers
- Stanford University · 2 papers
- University of Illinois, Urbana Champaign · 2 papers
- Max Planck Institute for Informatics · 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
- Hand-object interaction generation (116 papers)
- Scene-aware human motion synthesis (82 papers)
- Learning physics from video (74 papers)
Elsewhere in the topic
- Learning 3D reconstruction from images (196 papers)
- Diffusion-based novel view synthesis (156 papers)
- Semantic scene completion (153 papers)
- Generative indoor scene synthesis (131 papers)
- Feed-forward 3D asset generation and editing (116 papers)
- Score distillation for text-to-3D (106 papers)
- Open-vocabulary 3D scene understanding (95 papers)
- 2D-to-3D pretrained representations (94 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).