OccuFly: A 3D Vision Benchmark for Semantic Scene Completion from the Aerial Perspective
Published:Dec 25, 2025 05:00
•1 min read
•ArXiv Vision
Analysis
This paper introduces OccuFly, a novel benchmark dataset for semantic scene completion (SSC) from an aerial perspective, addressing a gap in existing research that primarily focuses on terrestrial environments. The key innovation lies in its camera-based data generation framework, which circumvents the limitations of LiDAR sensors on UAVs. By providing a diverse dataset captured across different seasons and environments, OccuFly enables researchers to develop and evaluate SSC algorithms specifically tailored for aerial applications. The automated label transfer method significantly reduces the manual annotation effort, making the creation of large-scale datasets more feasible. This benchmark has the potential to accelerate progress in areas such as autonomous flight, urban planning, and environmental monitoring.
Key Takeaways
Reference
“Semantic Scene Completion (SSC) is crucial for 3D perception in mobile robotics, as it enables holistic scene understanding by jointly estimating dense volumetric occupancy and per-voxel semantics.”