Research map / 3D vision / Depth and stereo
AerialMetric: Benchmarking and Adapting UAV Monocular Metric Depth Estimation in the Real World
Zhongqiang Song, Guanying Chen, Yuqi Zhang, Yin Zou, Chuanyu Fu, Zhiyuan Yuan, Chuan Huang, Shuguang Cui, Xiaochun Cao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| fine-tune MoGe2 on AerialMetric | MoGe2 · AerialMetric-Oblique City (w/o GT intrinsics) | delta1: 5.1 -> 89.3 | Improved |
| fine-tune MoGe2 on AerialMetric | MoGe2 · AerialMetric-Wild (0-400m) | AbsRel: 55.26 -> 21.34 | Improved |
| fine-tune MoGe2 on AerialMetric | MoGe2 · AerialMetric-Wild (0-400m) | delta1: 9.5 -> 53.7 | Improved |
| provide GT camera intrinsics at test time | MoGe2 · AerialMetric-Oblique City | delta1: 5.1 -> 18.1 | Improved |
| fine-tune MoGe2 on AerialMetric | MoGe2 · KITTI and ETH3D (ground, w/o GT intrinsics) | delta1: KITTI 64.7 -> 68.7; ETH3D 88.8 -> 86.7 | Mixed |
| fine-tune MoGe2 on AerialMetric | MoGe2 · HAMMER (w/o GT intrinsics) | AbsRel: 26.0 -> 31.7 | Hurt |
| use full fine-tuning instead of LoRA r=96 (ablation) | MoGe2 · Ground-domain mean | delta1: 79.59 -> 56.70 | Hurt |
| use LoRA r=128 instead of r=96 (ablation) | MoGe2 LoRA · Ground-domain mean | delta1: 79.59 -> 69.40 | Hurt |
| update only scale head instead of LoRA r=96 (ablation) | MoGe2 · AerialMetric-Oblique | delta1: 84.40 -> 53.00 | Hurt |
| add AerialMetric-Synthetic to Oblique training data (ablation) | MoGe2 LoRA · AerialMetric-Decoupled (mean) | delta1: 77.45 -> 84.20 | Improved |
| add AerialMetric-Synthetic to Oblique training data (ablation) | MoGe2 LoRA · AerialMetric-Oblique (mean) | delta1: 87.4 -> 82.30 | Hurt |
| add small fraction of ground-domain data (ablation) | MoGe2 LoRA · ETH3D | delta1: 71.54 -> 86.67 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MoGe2 — MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Details (NeurIPS 2025) · better on 9 of 11 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- UAV3D: A Large-scale 3D Perception Benchmark for Unmanned Aerial Vehicles (NeurIPS 2024 · similarity 0.87)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.87)
- Seeing Across Skies and Streets: Feedforward 3D Reconstruction from Satellite, Drone, and Ground Images (NeurIPS 2026 · similarity 0.86)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.85)
- AerialMegaDepth: Learning Aerial-Ground Reconstruction and View Synthesis (CVPR 2025 · similarity 0.85)
- OrthoLoC: UAV 6-DoF Localization and Calibration Using Orthographic Geodata (NeurIPS 2025 · similarity 0.85)
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.
