Stpls3D provides synthetic and real-world 2D and 3D point cloud data for semantic and instance segmentation, published by meidachen for BMVC 2022. The dataset utilizes AirSim for synthetic generation alongside real-world photogrammetry captures to support 3D reconstruction benchmarks.
Use Cases
- Training semantic segmentation models using 3D point cloud coordinates
- Evaluating instance segmentation performance on urban photogrammetry scenes
- Benchmarking 3D reconstruction algorithms using synthetic AirSim ground truth
Strengths
- Peer-reviewed BMVC 2022 Oral publication
- Includes both synthetic (AirSim) and real-world data
- Supports both semantic and instance segmentation labels
Limitations
- Unknown total record count and file size in provided metadata
- Potential domain shift between synthetic AirSim data and real-world captures
Provenance
- Source
- BMVC 2022 Oral publication by meidachen
- Collection Method
- Synthetic generation via AirSim and real-world photogrammetry
- Freshness
- Last updated December 2025.