# Learning the Target Priors Before Image Translation: A Decoupled Training Paradigm for Cross-Modal Image Translation in Remote Sensing

arXiv: https://arxiv.org/abs/2608.28517 · HTML: https://mlsanity.com/paper/2608.28517/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 44 references that are in the map.
- First posted on arXiv 28 Aug 2026.

## Builds on (most cited first, 25 of 25)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Image-to-Image Translation with Conditional Adversarial Networks](https://mlsanity.com/paper/1611.07004/) (2016)
- [Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks](https://mlsanity.com/paper/1703.10593/) (2017)
- [Reliable Fidelity and Diversity Metrics for Generative Models](https://mlsanity.com/paper/2002.09797/) (2020)
- [OminiControl: Minimal and Universal Control for Diffusion Transformer](https://mlsanity.com/paper/2411.15098/) (2024)
- [Rethinking FID: Towards a Better Evaluation Metric for Image Generation](https://mlsanity.com/paper/2401.09603/) (2023)
- [Toward Multimodal Image-to-Image Translation](https://mlsanity.com/paper/1711.11586/) (2017)
- [BBDM: Image-to-image Translation with Brownian Bridge Diffusion Models](https://mlsanity.com/paper/2205.07680/) (2022)
- [I2SB: Image-to-Image Schr"odinger Bridge](https://mlsanity.com/paper/2302.05872/) (2023)
- [DiffusionSat: A Generative Foundation Model for Satellite Imagery](https://mlsanity.com/paper/2312.03606/) (2023)
- [Ctrl-Adapter: An Efficient and Versatile Framework for Adapting Diverse Controls to Any Diffusion Model](https://mlsanity.com/paper/2404.09967/) (2024)
- [CRS-Diff: Controllable Remote Sensing Image Generation with Diffusion Model](https://mlsanity.com/paper/2403.11614/) (2024)
- [Text2Earth: Unlocking Text-driven Remote Sensing Image Generation with a Global-Scale Dataset and a Foundation Model](https://mlsanity.com/paper/2501.00895/) (2025)
- [SpaceNet 6: Multi-Sensor All Weather Mapping Dataset](https://mlsanity.com/paper/2004.06500/) (2020)
- [Denoising Likelihood Score Matching for Conditional Score-based Data Generation](https://mlsanity.com/paper/2203.14206/) (2022)
- [MetaEarth: A Generative Foundation Model for Global-Scale Remote Sensing Image Generation](https://mlsanity.com/paper/2405.13570/) (2024)
- [The QXS-SAROPT Dataset for Deep Learning in SAR-Optical Data Fusion](https://mlsanity.com/paper/2103.08259/) (2021)
- [The Prism Hypothesis: Harmonizing Semantic and Pixel Representations via Unified Autoencoding](https://mlsanity.com/paper/2512.19693/) (2025)
- [Generative Artificial Intelligence Meets Synthetic Aperture Radar: A Survey](https://mlsanity.com/paper/2411.05027/) (2024)
- [C-DiffSET: Leveraging Latent Diffusion for SAR-to-EO Image Translation with Confidence-Guided Reliable Object Generation](https://mlsanity.com/paper/2411.10788/) (2024)
- [Any2Any: Unified Arbitrary Modality Translation for Remote Sensing](https://mlsanity.com/paper/2603.04114/) (2026)
- [UniCon: Unidirectional Information Flow for Effective Control of Large-Scale Diffusion Models](https://mlsanity.com/paper/2503.17221/) (2025)
- [Adaptive Domain Shift in Diffusion Models for Cross-Modality Image Translation](https://mlsanity.com/paper/2601.18623/) (2026)

## Abstract (arXiv)

Cross-modal image translation in remote sensing must preserve source-observed content while matching the target-domain distribution. Existing methods jointly learn the target prior and cross-modal dependence from scarce paired data, overlooking a key asymmetry: only the latter intrinsically requires cross-modal correspondence. We formalize this distinction through conditional-score and denoising-risk analyses and propose Learning the Target Priors Before Image Translation (LTP-BIT), a prior-first paradigm that decouples the two learning tasks. LTP-BIT first learns a target-domain generative prior from large-scale unpaired imagery, then retains the pretrained backbone weights and learns source-conditioned control through P-DART, a parameter-efficient dual-stream architecture. Controlled experiments show that prior matching and scaling primarily improve target-domain realism, whereas instance fidelity relies more strongly on conditional adaptation. LTP-BIT achieves state-of-the-art performance across SAR-to-RGB and NIR-to-RGB benchmarks using only 9.81% task-specific parameters. On QXS-SAROPT, it retains near-full-data instance fidelity with only 25% of the paired samples.