Research map / 3D vision / 3D generation
Geometry-Guided Street-View Panorama Synthesis From Satellite Imagery
Yujiao Shi, Dylan Campbell, Xin Yu, Hongdong Li
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 S2SP module (satellite-to-street projection) to the generator (ablation) | Pix2Pix-style generator · CVACT (Aligned) | SSIM: 0.4050 -> 0.4212 | Improved |
| add S2SP module (satellite-to-street projection) to the generator (ablation) | Pix2Pix-style generator · CVUSA | SSIM: 0.3197 -> 0.3408 | Improved |
| add learned height estimation (vs zero height projection) (ablation) | S2SP · CVUSA | SSIM: 0.3301 -> 0.3408 | Improved |
| add geometry projection block instead of soft-argmax height concatenation (ablation) | S2SP · CVACT (Aligned) | RMSE: 49.47 -> 48.23 | Improved |
| add perceptual loss (ablation) | S2SP · CVACT (Aligned) | RMSE: 50.96 -> 48.23 | Improved |
| add L1 loss (ablation) | S2SP · CVUSA | SSIM: 0.1512 -> 0.3408 | Improved |
| add L1 loss (ablation) | S2SP · CVACT (Aligned) | SSIM: 0.4174 -> 0.4212 | No clear effect |
| use deep generator instead of handcrafted polar transform (ablation) | S2SP · CVACT (Aligned) | RMSE: 82.08 -> 48.23 | Improved |
| increase MPI planes N from 1 to 64 (ablation) | S2SP · CVUSA | SSIM: 0.3301 -> 0.3408 | Improved |
| use N=16 planes instead of 64 (ablation) | S2SP · CVACT (Aligned) | RMSE: 47.89 -> 48.23 | No clear effect |
| train on aligned instead of unaligned pairs | S2SP · CVACT (Aligned) | RMSE: 48.37 -> 48.23 | No clear effect |
| train on aligned instead of unaligned pairs, test on unaligned set | S2SP · CVACT (Unaligned test) | RMSE: 49.82 -> 50.66 | Hurt |
Closest papers by meaning
- Geometry-Aware Satellite-to-Ground Image Synthesis for Urban Areas (CVPR 2020 · similarity 0.92)
- Seeing Through Satellite Images at Street Views (TPAMI 2026 · similarity 0.90)
- Coming Down to Earth: Satellite-to-Street View Synthesis for Geo-Localization (CVPR 2021 · similarity 0.89)
- Controllable Satellite-to-Street-View Synthesis with Precise Pose Alignment and Zero-Shot Environmental Control (ICLR 2025 · similarity 0.87)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.84)
- HORIZON: High-Resolution Semantically Controlled Panorama Synthesis (AAAI 2024 · 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.
