Research map / 3D vision / Face and shape
Full-body avatar reconstruction
Reconstructs digital humans and articulated objects from monocular or single-view inputs with diffusion or disentangled modeling.
78 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Face and shape, keyword cluster “avatar · human · body”
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.
28 papers in 2023–24 and 22 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 7.7% to 6.0%.
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 Human Body Correspondences Using Convolutional Networks (CVPR 2016 · 187 citations)
- 2017 · Dynamic FAUST: Registering Human Bodies in Motion (CVPR 2017 · 378 citations)
- 2018 · Total Capture: A 3D Deformation Model for Tracking Faces, Hands, and Bodies (CVPR 2018 · 548 citations)
- 2019 · Gait Recognition via Disentangled Representation Learning (CVPR 2019 · 292 citations)
- 2020 · HUMBI: A Large Multiview Dataset of Human Body Expressions (CVPR 2020 · 87 citations)
- 2021 · Neural Body: Implicit Neural Representations With Structured Latent Codes for Novel View Synthesis of Dynamic Humans (CVPR 2021 · 633 citations)
- 2022 · Gait Recognition in the Wild With Dense 3D Representations and a Benchmark (CVPR 2022 · 256 citations)
- 2023 · High-Fidelity 3D Human Digitization From Single 2K Resolution Images (CVPR 2023 · 50 citations)
- 2024 · GART: Gaussian Articulated Template Models (CVPR 2024 · 99 citations)
- 2025 · PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing (CVPR 2025 · 19 citations)
- 2026 · Densemarks: Learning Canonical Embeddings for Human Heads Images via Point Tracks (ICLR 2026 · 0 citations)
Earliest papers here
- Dense Human Body Correspondences Using Convolutional Networks (CVPR 2016 · 187 citations)
- Dynamic FAUST: Registering Human Bodies in Motion (CVPR 2017 · 378 citations)
- Total Capture: A 3D Deformation Model for Tracking Faces, Hands, and Bodies (CVPR 2018 · 548 citations)
What has been tried
66 papers here report 608 measured changes (a component added, removed or swapped, compared with a baseline): helped 81%, no clear effect 2%, hurt 15%, 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 a per-frame table instead of an MLP for latent bone transformations | UBC-Fashion · PSNR | 25.65 -> 25.96 | Helped | GART: Gaussian Articulated Template Models (CVPR 2024 · 99 citations) |
| add skinned body mesh guidance | CustomHumans · f-Score | 33.244 -> 37.029 | Helped | SiTH: Single-view Textured Human Reconstruction with Image-Conditioned Diffusion (CVPR 2024 · 52 citations) |
| add rigidity loss | Unbiased4D synthetic cactus · PSNR | 20.73 -> 22.348 | Helped | Neural Parametric Gaussians for Monocular Non-Rigid Object Reconstruction (CVPR 2024 · 24 citations) |
| add SMPL-X conditioning to the diffusion model | CAPE-NFP subset · Chamfer distance | 1.4370 -> 1.0938 | Helped | PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing (CVPR 2025 · 19 citations) |
| Use a spatially varying linear lighting model rather than non-linear activations | Fully lit · PSNR | 17.50 -> 20.61 | Helped | URHand: Universal Relightable Hands (CVPR 2024 · 17 citations) |
| use best of 10 samples instead of 1 | Test set of [5] · Albedo back PSNR | 19.77 -> 21.24 | Helped | DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humans (CVPR 2024 · 15 citations) |
| add pose-dependent deformations | In-shape testset · Chamfer distance (mm) | 10.32 -> 9.37 | Helped | VAREN: Very Accurate and Realistic Equine Network (CVPR 2024 · 14 citations) |
| Use a displacement field instead of QRBS | USB · F(10%) | 59.5 -> 78.6 | Hurt | REACTO: Reconstructing Articulated Objects from a Single Video (CVPR 2024 · 13 citations) |
| add texture-consistent back-view synthesis | THuman2.0 · PSNR | 20.188 -> 21.27 | Helped | ConTex-Human: Free-View Rendering of Human from a Single Image with Texture-Consistent Synthesis (CVPR 2024 · 12 citations) |
| Add progressive SAM prompting to layer-wise volume rendering | Hi4D · V-IoU | 0.773 -> 0.786 | Helped | MultiPly: Reconstruction of Multiple People from Monocular Video in the Wild (CVPR 2024 · 12 citations) |
| Add HR features and VFE to LR-only features | RenderPeople and THuman2.0 · CD | 2.653 -> 2.075; 1.998 -> 0.913 | Helped | ANIM: Accurate Neural Implicit Model for Human Reconstruction from a single RGB-D Image (CVPR 2024 · 11 citations) |
| use backward rather than forward deformations | DPHM-Kinect · ℓ2 | 1.367 -> 1.146 | Helped | DPHMs: Diffusion Parametric Head Models for Depth-based Tracking (CVPR 2024 · 7 citations) |
What reviewers will expect
Datasets and baselines named in the 10 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CustomHumans · 4 of 10 papers
- PeopleSnapshot · 2 of 10 papers
- THuman2.1 · 2 of 10 papers
- CAPE · 2 of 10 papers
- ZJU-MoCap · 2 of 10 papers
- AIST++ · 2 of 10 papers
- THuman3.0 · 2 of 10 papers
- X-Humans · 2 of 10 papers
Baselines
- SIFU · 5 of 10 papers
- SiTH · 5 of 10 papers
- ECON · 4 of 10 papers
- GTA · 4 of 10 papers
- ICON · 3 of 10 papers
- TeCH · 3 of 10 papers
- PIFu · 2 of 10 papers
- PaMIR · 2 of 10 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 50 papers; from OpenAlex and the papers' affiliations).
- Nanyang Technological University · 5 papers
- Shanghai Jiao Tong University · 2 papers
- A*STAR · 2 papers
- Google · 2 papers
- University of Pennsylvania · 2 papers
- Wuhan University · 2 papers
- Peking University · 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
- Gaussian and implicit head avatars (115 papers)
- Clothed human modeling (60 papers)
- Generative textures for 3D humans (27 papers)
Elsewhere in the topic
- Neural implicit surface reconstruction (131 papers)
- Textured mesh generation (111 papers)
- Primitive-based shape abstraction (99 papers)
- Single-view 3D reconstruction (96 papers)
- Face reflectance and relightable rendering (85 papers)
- 3D morphable face model regression (80 papers)
- Articulated and non-rigid shape learning (75 papers)
- 3D shape generative modeling (74 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).