Research map / 3D vision / 3D generation
3D perception for autonomous driving
Targets lanes, mapping, pretraining and rendering for driving, with large benchmarks.
73 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in 3D generation, keyword cluster “semantic · occupancy · driving”
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.
21 papers in 2023–24 and 38 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 4.6% to 4.3%.
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
- 2018 · Viewpoint-Aware Attentive Multi-View Inference for Vehicle Re-Identification (CVPR 2018 · 300 citations)
- 2019 · Argoverse: 3D Tracking and Forecasting With Rich Maps (CVPR 2019 · 1,487 citations)
- 2020 · Scalability in Perception for Autonomous Driving: Waymo Open Dataset (CVPR 2020 · 3,252 citations)
- 2021 · Projecting Your View Attentively: Monocular Road Scene Layout Estimation via Cross-View Transformation (CVPR 2021 · 88 citations)
- 2022 · Rope3D: The Roadside Perception Dataset for Autonomous Driving and Monocular 3D Object Detection Task (CVPR 2022 · 154 citations)
- 2023 · UniSim: A Neural Closed-Loop Sensor Simulator (CVPR 2023 · 157 citations)
- 2024 · UniPAD: A Universal Pre-training Paradigm for Autonomous Driving (CVPR 2024 · 34 citations)
- 2025 · ReconDreamer: Crafting World Models for Driving Scene Reconstruction via Online Restoration (CVPR 2025 · 23 citations)
- 2026 · SparseCoop: Cooperative Perception with Kinematic-Grounded Queries (AAAI 2026 · 1 citation)
Earliest papers here
- Viewpoint-Aware Attentive Multi-View Inference for Vehicle Re-Identification (CVPR 2018 · 300 citations)
- Argoverse: 3D Tracking and Forecasting With Rich Maps (CVPR 2019 · 1,487 citations)
- ApolloCar3D: A Large 3D Car Instance Understanding Benchmark for Autonomous Driving (CVPR 2019 · 195 citations)
What has been tried
49 papers here report 531 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 3%, hurt 18%, 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 |
|---|---|---|---|---|
| vary the camera input mask ratio | nuScenes val, camera detection · NDS | 31.9; 32.9; 32.3; 32.1; 31.4 | Mixed | UniPAD: A Universal Pre-training Paradigm for Autonomous Driving (CVPR 2024 · 34 citations) |
| add the novel continuous lane representation | OpenLane300 · F1 | 62.9 -> 65.0 | Helped | LaneCPP: Continuous 3D Lane Detection using Physical Priors (CVPR 2024 · 31 citations) |
| add masks during DriveDreamer-2 fine-tuning | Waymo novel trajectories, average · NTA-IoU | 0.511 -> 0.517 | Helped | ReconDreamer: Crafting World Models for Driving Scene Reconstruction via Online Restoration (CVPR 2025 · 23 citations) |
| Add rendered semantic and depth maps (Gaussian-based joint rendering) as occupancy conditions in place of spatial-temporal attention | NuScenes video generation · FVD | FVD 110.87 -> 70.52 (about 34.66% and 31.15% gains from semantic/depth maps) | Helped | UniScene: Unified Occupancy-centric Driving Scene Generation (CVPR 2025 · 22 citations) |
| add ground-truth language cues to the Executor | Planning · L2 (m) | 1.03 -> 0.94 | Helped | DME-Driver: Integrating Human Decision Logic and 3D Scene Perception in Autonomous Driving (AAAI 2025 · 21 citations) |
| add PAN to the baseline | OpenLane validation · F1 | 55.3 -> 59.3 | Helped | PVALane: Prior-Guided 3D Lane Detection with View-Agnostic Feature Alignment (AAAI 2024 · 19 citations) |
| Use Local MVR instead of Global MVR | nuScenes, Front camera missing · mAP | 0.3160 -> 0.3263 | Helped | M-BEV: Masked BEV Perception for Robust Autonomous Driving (AAAI 2024 · 14 citations) |
| Add PointNMS | OpenLane300 · F1-score | 66.6 -> 69.2 | Helped | GLane3D: Detecting Lanes with Graph of 3D Keypoints (CVPR 2025 · 7 citations) |
| Add EP-head | OpenLane, Lane-IoU=0.75 · F1 | 61.9 -> 63.1 | Helped | Rethinking Lanes and Points in Complex Scenarios for Monocular 3D Lane Detection (CVPR 2025 · 7 citations) |
| add intra-modal distillation to CaDDN | KITTI val, Car · AP3D | 24.35 -> 31.11; 16.16 -> 20.24; 13.44 -> 16.91 | Helped | MonoTAKD: Teaching Assistant Knowledge Distillation for Monocular 3D Object Detection (CVPR 2025 · 6 citations) |
| add sparse feature inpainting | Occ3D-nuScenes · mIoU | 21.9 -> 23.8 | Helped | Regulating Intermediate 3D Features for Vision-Centric Autonomous Driving (AAAI 2024 · 6 citations) |
| increase octree depth while keeping the smallest voxel resolution fixed | VKITTI2 · L2 Loss | 712.09 -> 535.32 -> 420.03 -> 339.68 -> 317.01 | Helped | Swift-Mapping: Online Neural Implicit Dense Mapping in Urban Scenes (AAAI 2024 · 5 citations) |
What reviewers will expect
Datasets and baselines named in the 19 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- nuScenes · 9 of 19 papers
- OpenLane · 2 of 19 papers
- Waymo Open Dataset · 2 of 19 papers
- Argoverse2 · 2 of 19 papers
Baselines
- UniAD · 3 of 19 papers
- BEVFormer · 3 of 19 papers
- MagicDrive · 2 of 19 papers
- Panacea · 2 of 19 papers
- PersFormer · 2 of 19 papers
- BEV-LaneDet · 2 of 19 papers
- Anchor3DLane · 2 of 19 papers
- LATR · 2 of 19 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 59 papers; from OpenAlex and the papers' affiliations).
- Tsinghua University · 4 papers
- Zhejiang University · 3 papers
- University of Macau · 3 papers
- Shanghai Jiao Tong University · 3 papers
- Chinese Academy of Sciences · 3 papers
- Chinese academy of science · 2 papers
- University of Southern California · 2 papers
- Shanghai AI Laboratory · 2 papers
- Waabi · 2 papers
- University of Toronto · 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
- Learning 3D reconstruction from images (196 papers)
- Semantic scene completion (153 papers)
- 2D-to-3D pretrained representations (94 papers)
- Vision-based 3D occupancy prediction (50 papers)
Elsewhere in the topic
- Diffusion-based novel view synthesis (156 papers)
- Generative indoor scene synthesis (131 papers)
- Hand-object interaction generation (116 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)
- Monocular 3D object reconstruction (94 papers)
- Scene-aware human motion synthesis (82 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).