Research map / Image generation / Autoregressive and latent generation
Object Fidelity Diffusion for Remote Sensing Image Generation
Ziqi Ye, Shuran Ma, Jie Yang, Xiaoyi Yang, Yi Yang, Ziyang Gong, Xue Yang, Haipeng Wang
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 enhanced shape generation module (ESGM) (ablation) | OF-Diff · DIOR | FID: 42.59 -> 24.87 | Improved |
| add ESGM (ablation) | OF-Diff · DIOR | YOLOScore: 41.20 -> 55.08 | Improved |
| add online-distillation consistency loss L_c (ablation) | OF-Diff · DIOR | FID: 42.59 -> 36.25 | Improved |
| add DDPO fine-tuning to full model with ESGM and L_c (ablation) | OF-Diff · DIOR | mAP50: 54.31 -> 54.44 | Improved |
| add DDPO fine-tuning (ablation) | OF-Diff with ESGM and L_c · DIOR | FID: 24.98 -> 24.92 | No clear effect |
| add DDPO alone without ESGM (ablation) | OF-Diff · DIOR | mAP50: 52.13 -> 53.41 | Improved |
| add captions as conditioning input (ablation) | OF-Diff · DIOR | Delta mAP50: -1.15 | Hurt |
| add captions as conditioning input (ablation) | OF-Diff · DIOR aesthetic preference | preference rate: Q2 caption 11.21 vs 8.79; Q1 realism 6.57 vs 13.43 | Mixed |
| replace real training data with only generated data (ablation) | OF-Diff · DIOR | mAP: 52.84 -> 45.67 | Hurt |
| add generated data on top of real data (up to 300%) (ablation) | OF-Diff · DIOR | mAP: 52.84 -> 54.82 | Improved |
| use OF-Diff instead of CC-Diff on HRSC2016 for CMMD/CAS/YOLOScore | OF-Diff · HRSC2016 | YOLOScore: 32.42 -> 30.97 | Hurt |
| use OF-Diff instead of CC-Diff on DIOR for CAS | OF-Diff · DIOR | CAS: 82.61 -> 82.55 | No clear effect |
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.
- LayoutDiffusion — LayoutDiffusion: Controllable Diffusion Model for Layout-to-Image Generation (CVPR 2023) · better on 2 of 2 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
- DiffusionSat: A Generative Foundation Model for Satellite Imagery (ICLR 2024 · similarity 0.87)
- Building Bridges across Spatial and Temporal Resolutions: Reference-Based Super-Resolution via Change Priors and Conditional Diffusion Model (CVPR 2024 · similarity 0.86)
- SSDiff: Spatial-spectral Integrated Diffusion Model for Remote Sensing Pansharpening (NeurIPS 2024 · similarity 0.86)
- MMM-RS: A Multi-modal, Multi-GSD, Multi-scene Remote Sensing Dataset and Benchmark for Text-to-Image Generation (NeurIPS 2024 · similarity 0.85)
- Diffusing Differentiable Representations (NeurIPS 2024 · similarity 0.85)
- The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noise (ICLR 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.
