Physics-Informed Diffusion Models for SAR Ship Wake Generation from Text Prompts
This work addresses data scarcity for maritime SAR analysis, offering a faster and more efficient method for generating ship wake images, though it is incremental as it builds on existing diffusion models and simulation techniques.
The paper tackles the problem of limited annotated data for ship wake detection in SAR imagery by using a diffusion model trained on physics-based simulation data to generate realistic Kelvin wake patterns from text prompts, achieving significantly faster inference than the simulator.
Detecting ship presence via wake signatures in SAR imagery is attracting considerable research interest, but limited annotated data availability poses significant challenges for supervised learning. Physics-based simulations are commonly used to address this data scarcity, although they are slow and constrain end-to-end learning. In this work, we explore a new direction for more efficient and end-to-end SAR ship wake simulation using a diffusion model trained on data generated by a physics-based simulator. The training dataset is built by pairing images produced by the simulator with text prompts derived from simulation parameters. Experimental result show that the model generates realistic Kelvin wake patterns and achieves significantly faster inference than the physics-based simulator. These results highlight the potential of diffusion models for fast and controllable wake image generation, opening new possibilities for end-to-end downstream tasks in maritime SAR analysis.