AI Paper Atlas

Research map / Point clouds / Segmentation and detection

UAVScenes: A Multi-Modal Dataset for UAVs

Sijie Wang, Siqi Li, Yawei Zhang, Shangshu Yu, Shenghai Yuan, Rui She, Quanjiang Guo, JinXuan Zheng, Ong Kang Howe, Leonrich Chandra, Shrivarshann Srijeyan, Aditya Sivadas and 7 more

ICCV 2025 · Point clouds › Segmentation and detection · approach: LiDAR segmentation for driving · 22 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use a Transformer backbone (DeiT3-m) instead of CNN (ResNet-50)UperNet segmentation · UAVScenes image semantic segmentationmIoU: 61.3 -> 68.3Improved
use ViT-t instead of ConvNext-tUperNet segmentation · UAVScenes image semantic segmentationmIoU: 55.3 -> 62.8Improved
use PTv2 (11M) instead of MinkUNet (38M)LiDAR segmentation · UAVScenes LiDAR semantic segmentationmIoU: 32.7 -> 33.2Improved
use fusion (MinkLoc++) instead of LiDAR only (MinkLoc3D V2)place recognition · UAVScenes place recognitionRecall@1: 42.8 -> 47.1Improved
use DINOv2 foundation backbone (SALAD) instead of ResNet-18 (GeM)camera place recognition · UAVScenes place recognitionRecall@1: 42.1 -> 67.1Improved
use BEV projection (BEVPlace) instead of voxels (MinkLoc3D)LiDAR place recognition · UAVScenes place recognitionRecall@1: 41.9 -> 32.6Hurt
use fusion UMF instead of single-modal MinkLoc3Dplace recognition · UAVScenes place recognitionRecall@1: 41.9 -> 40.1No clear effect
use absolute pose regression (RobustLoc) instead of scene coordinate regression (ACE)6-DoF localization · UAVScenes localization averagemedian position error (m): 132.9 -> 6.1Improved
use 3DGS instead of Instant-NGPnovel view synthesis · UAVScenes NVS AMtownPSNR: 20.09 -> 22.95Improved
use Metric3D V2 (ViT-l) instead of Depth Anything V2 (ViT-l)zero-shot depth · UAVScenes depthAbsRel: 1.517 -> 0.540Improved
use single-step models instead of diffusion models (Marigold)zero-shot depth · UAVScenes depthAbsRel: 0.994 -> 0.540Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy