# ZVeC: A Zero-Shot Framework for Instance-Level Vehicle Extraction and Generative Point Cloud Completion

arXiv: https://arxiv.org/abs/2609.24825 · HTML: https://mlsanity.com/paper/2609.24825/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 23 references that are in the map.
- First posted on arXiv 21 Sep 2026.

## Builds on (most cited first, 22 of 22)

- [OpenScene: 3D Scene Understanding with Open Vocabularies](https://mlsanity.com/paper/2211.15654/) (2022)
- [Deep Learning for 3D Point Clouds: A Survey](https://mlsanity.com/paper/1912.12033/) (2019)
- [PCN: Point Completion Network](https://mlsanity.com/paper/1808.00671/) (2018)
- [Mask3D: Mask Transformer for 3D Semantic Instance Segmentation](https://mlsanity.com/paper/2210.03105/) (2022)
- [PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers](https://mlsanity.com/paper/2108.08839/) (2021)
- [SAM 3D: 3Dfy Anything in Images](https://mlsanity.com/paper/2511.16624/) (2025)
- [3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans](https://mlsanity.com/paper/1812.07003/) (2018)
- [SoftGroup for 3D Instance Segmentation on Point Clouds](https://mlsanity.com/paper/2203.01509/) (2022)
- [OccuSeg: Occupancy-aware 3D Instance Segmentation](https://mlsanity.com/paper/2003.06537/) (2020)
- [GRNet: Gridding Residual Network for Dense Point Cloud Completion](https://mlsanity.com/paper/2006.03761/) (2020)
- [SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer](https://mlsanity.com/paper/2108.04444/) (2021)
- [Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds](https://mlsanity.com/paper/1906.01140/) (2019)
- [GSPN: Generative Shape Proposal Network for 3D Instance Segmentation in Point Cloud](https://mlsanity.com/paper/1812.03320/) (2018)
- [Hierarchical Aggregation for 3D Instance Segmentation](https://mlsanity.com/paper/2108.02350/) (2021)
- [Deep Learning for Image and Point Cloud Fusion in Autonomous Driving: A Review](https://mlsanity.com/paper/2004.05224/) (2020)
- [PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things](https://mlsanity.com/paper/1903.01177/) (2019)
- [SAI3D: Segment Any Instance in 3D Scenes](https://mlsanity.com/paper/2312.11557/) (2023)
- [SeedFormer: Patch Seeds based Point Cloud Completion with Upsample Transformer](https://mlsanity.com/paper/2207.10315/) (2022)
- [AdaPoinTr: Diverse Point Cloud Completion with Adaptive Geometry-Aware Transformers](https://mlsanity.com/paper/2301.04545/) (2023)
- [AVP-SLAM: Semantic Visual Mapping and Localization for Autonomous Vehicles in the Parking Lot](https://mlsanity.com/paper/2007.01813/) (2020)
- [GenPC: Zero-shot Point Cloud Completion via 3D Generative Priors](https://mlsanity.com/paper/2502.19896/) (2025)
- [Towards Scalable Spatial Intelligence via 2D-to-3D Data Lifting](https://mlsanity.com/paper/2507.18678/) (2025)

## Abstract (arXiv)

LiDAR point clouds acquired in underground environments exhibit severe geometric incompleteness due to occlusions and limited sensor viewpoints, making reliable point cloud completion challenging without large supervised datasets. We propose ZVeC, a zero-shot, instance-driven framework that reformulates scene-level completion as compositional object-level reconstruction. By decomposing a scene into semantic object instances, ZVeC reduces reconstruction ambiguity in cluttered environments while eliminating the need for scenario-specific training. Each segmented vehicle is completed independently using a depth- and 3D Gaussian-conditioned diffusion model that exploits generalized geometric priors before the reconstructed instances are recomposed into the original scene. To evaluate our approach, we construct a real-world dense LiDAR benchmark of underground parking environments. Experimental results demonstrate consistent improvements over representative scene-level baselines in both quantitative metrics and visual quality. The completed point cloud differs substantially from the measured input (average KL divergence ~ 2.1), yet reducing the input to only 1% of the original LiDAR measurements changes the completed reconstruction only marginally (KL divergence < 0.50). This demonstrates that ZVeC produces geometrically consistent completions even under extreme input sparsity.