Research map / 3D vision / 3D generation
3D-LENS: A 3D Lifting-based Elevated Novel-view Synthesis method for Single-View Aerial-Ground Re-Identification
William Grolleau, Astrid Sabourin, Guillaume Lapouge, Catherine Achard
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 |
|---|---|---|---|
| add novel-view synthesis alone (raw rendered views) (ablation) | ViT baseline · AG-ReID Ground training (A->G / G->A) | mAP: 46.9 -> 43.7 (A->G), 49.8 -> 48.3 (G->A) | Hurt |
| add background compositing to novel views (ablation) | Nv only · AG-ReID Ground training (A->G / G->A) | mAP: 43.7 -> 43.5 (A->G), 48.3 -> 50.2 (G->A) | Mixed |
| add style transfer to novel views (ablation) | Nv only · AG-ReID Ground training (A->G / G->A) | mAP: 43.7 -> 48.6 (A->G), 48.3 -> 51.0 (G->A) | Improved |
| add background compositing on top of style transfer (ablation) | Nv+St · AG-ReID Ground training (A->G / G->A) | mAP: 48.6 -> 50.9 (A->G), 51.0 -> 51.6 (G->A) | Improved |
| add balanced sampling to novel views (ablation) | Nv only · AG-ReID Ground training (A->G / G->A) | mAP: 43.7 -> 47.0 (A->G), 48.3 -> 51.7 (G->A) | Improved |
| add balanced sampling to compositing plus style transfer (ablation) | Nv+Ba+St · AG-ReID Ground training (A->G / G->A) | mAP: 50.9 -> 51.2 (A->G), 51.6 -> 52.8 (G->A) | Improved |
| add domain classification loss (ablation) | Nv+Ba+St · AG-ReID Ground training (A->G / G->A) | mAP: 50.9 -> 51.4 (A->G), 51.6 -> 52.8 (G->A) | Improved |
| add curriculum scheduler (ablation) | Nv+Ba+St · AG-ReID Ground training (A->G / G->A) | mAP: 50.9 -> 51.7 (A->G), 51.6 -> 53.1 (G->A) | Improved |
| add balanced sampling, domain loss and curriculum together (ablation) | Nv+Ba+St · AG-ReID Ground training (A->G / G->A) | mAP: 50.9 -> 53.3 (A->G), 51.6 -> 55.1 (G->A) | Improved |
| full 3D-LENS over ViT baseline | ViT · AG-ReID Ground training (A->G / G->A) | mAP: 46.9 -> 53.3 (A->G), 49.8 -> 55.1 (G->A) | Improved |
| 3D-LENS versus DCAC under Aerial training | DCAC · AG-ReID (Aerial training, G->A) | mAP: 35.2 -> 49.5 | Improved |
| 3D-LENS versus PASS under Aerial training | PASS · AG-ReID.v2 (Aerial training, C->A) | mAP: 16.6 -> 37.8 | Improved |
Closest papers by meaning
- AerialMegaDepth: Learning Aerial-Ground Reconstruction and View Synthesis (CVPR 2025 · similarity 0.88)
- Seeing Across Skies and Streets: Feedforward 3D Reconstruction from Satellite, Drone, and Ground Images (NeurIPS 2026 · similarity 0.85)
- Multi-Modal Aerial-Ground Cross-View Place Recognition with Neural ODEs (CVPR 2025 · similarity 0.85)
- SeCap: Self-Calibrating and Adaptive Prompts for Cross-view Person Re-Identification in Aerial-Ground Networks (CVPR 2025 · similarity 0.85)
- View-decoupled Transformer for Person Re-identification under Aerial-ground Camera Network (CVPR 2024 · similarity 0.85)
- GSAlign: Geometric and Semantic Alignment Network for Aerial-Ground Person Re-Identification (NeurIPS 2025 · similarity 0.84)
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.
