Research map / 3D vision / Face and shape
Learned mesh and geometric representations
Designs neural networks that produce meshes and fields with topology, analytic or vector-field representations.
73 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Face and shape, keyword cluster “surface · mesh · implicit”
State History Tried 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 15 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 4.1% to 4.1%.
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 · FPNN: Field Probing Neural Networks for 3D Data (NeurIPS 2016 · 250 citations)
- 2017 · 3DMatch: Learning Local Geometric Descriptors From RGB-D Reconstructions (CVPR 2017 · 1,155 citations)
- 2018 · Neural 3D Mesh Renderer (CVPR 2018 · 1,016 citations)
- 2019 · Occupancy Networks: Learning 3D Reconstruction in Function Space (CVPR 2019 · 2,830 citations)
- 2020 · Differentiable Volumetric Rendering: Learning Implicit 3D Representations Without 3D Supervision (CVPR 2020 · 880 citations)
- 2021 · Automated Model Design and Benchmarking of Deep Learning Models for COVID-19 Detection with Chest CT Scans (AAAI 2021 · 45 citations)
- 2022 · Vox2Cortex: Fast Explicit Reconstruction of Cortical Surfaces From 3D MRI Scans With Geometric Deep Neural Networks (CVPR 2022 · 60 citations)
- 2023 · Neural Vector Fields: Implicit Representation by Explicit Learning (CVPR 2023 · 21 citations)
- 2024 · Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering (CVPR 2024 · 47 citations)
- 2025 · 3D Dental Model Segmentation with Geometrical Boundary Preserving (CVPR 2025 · 10 citations)
- 2026 · Unsupervised Multi-Parameter Inverse Solving for Reducing Ring Artifacts in 3D X-Ray CBCT (AAAI 2026 · 1 citation)
Earliest papers here
- FPNN: Field Probing Neural Networks for 3D Data (NeurIPS 2016 · 250 citations)
- Marr Revisited: 2D-3D Alignment via Surface Normal Prediction (CVPR 2016 · 236 citations)
- Geometry-Informed Material Recognition (CVPR 2016 · 56 citations)
What has been tried
61 papers here report 620 measured changes (a component added, removed or swapped, compared with a baseline): helped 70%, no clear effect 5%, hurt 22%, 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 |
|---|---|---|---|---|
| add test-time optimization to the pose-regression initialization | DeepFluoro · SMSR | 1% -> 87% | Helped | Intraoperative 2D/3D Image Registration via Differentiable X-ray Rendering (CVPR 2024 · 47 citations) |
| Increase MLP-Res width from 128 to 256 neurons at 2 layers | RGB map · PSNR | 45.34 -> 45.84 | Helped | NILUT: Conditional Neural Implicit 3D Lookup Tables for Image Enhancement (AAAI 2024 · 22 citations) |
| Use PoNQ instead of PoNQ-lite (four predicted points per cell rather than one pooled point) | Thingi30 · ECD | 0.055 -> 0.123 | Hurt | PoNQ: a Neural QEM-based Mesh Representation (CVPR 2024 · 20 citations) |
| add image features to the point network | MICCAI Challenge test set · mIoU | 95.119 -> 95.860 | Helped | 3D Dental Model Segmentation with Geometrical Boundary Preserving (CVPR 2025 · 10 citations) |
| add the learned sampler | FFHQ-512 · FID | 38.29 -> 4.97 | Helped | What You See is What You GAN: Rendering Every Pixel for High-Fidelity Geometry in 3D GANs (CVPR 2024 · 7 citations) |
| Add query point shifting | ABC-NEF · R5 | 29.2 -> 56.4 | Helped | 3D Neural Edge Reconstruction (CVPR 2024 · 6 citations) |
| use a low-dimensional polyrigid deformation instead of dense R^3 translations | DeepFluoro, Femur (L) · Dice | 0.97(0.02) -> 0.98(0.01) | Helped | PolyPose: Deformable 2D/3D Registration via Polyrigid Transformations (NeurIPS 2025 · 3 citations) |
| add a half-space map to quaternions | Point cloud alignment · test performance | Helped | Learning with 3D rotations, a hitchhiker's guide to SO(3) (ICML 2024 · 3 citations) | |
| Use the coarse-to-fine strategy instead of a single model | Developable surface optimization · dH−a | 0.608% -> 0.517% | Helped | GSO-Net: Grid Surface Optimization via Learning Geometric Constraints (AAAI 2024 · 1 citation) |
| increase SDF MLP width from 32 to 64 or 128 units per layer | Movi-B · Translation and rotation error | No clear effect | Learning rigid-body simulators over implicit shapes for large-scale scenes and vision (NeurIPS 2024 · 1 citation) | |
| Add thickness edges | Deforming Plate, next timestep · RMSE (×10³) | 17.420 -> 14.903 | Helped | Thickness-aware E(3)-Equivariant 3D Mesh Neural Networks (ICML 2025 · 0 citations) |
| Use feature multiplication instead of concatenation | 3D segmentation, 2D tomographic supervision · IOU | 0.880 -> 0.877 | No clear effect | Geometric Algebra Planes: Convex Implicit Neural Volumes (ICML 2025 · 0 citations) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 30 papers; from OpenAlex and the papers' affiliations).
- Inria · 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
- Neural implicit surface reconstruction (131 papers)
- Textured mesh generation (111 papers)
- Implicit neural representation design (52 papers)
- Pixel-aligned implicit human and hand reconstruction (39 papers)
Elsewhere in the topic
- Gaussian and implicit head avatars (115 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)
- Full-body avatar reconstruction (78 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).