Search papers, labs, and topics across Lattice.
The paper introduces RSGen, a plug-and-play framework for enhancing layout-to-image (L2I) synthesis in remote sensing by incorporating diverse edge guidance. RSGen uses a progressive enhancement strategy, first enriching edge map diversity through image-to-image generation from retrieved training instances, and then using these edge maps as conditioning for L2I models. Experiments demonstrate that RSGen significantly improves existing L2I models, achieving substantial gains in downstream object detection performance on the DOTA dataset.
Achieve nearly 10% mAP improvement in remote sensing object detection by guiding diffusion models with diverse edge maps derived from training data.
Diffusion models have significantly mitigated the impact of annotated data scarcity in remote sensing (RS). Although recent approaches have successfully harnessed these models to enable diverse and controllable Layout-to-Image (L2I) synthesis, they still suffer from limited fine-grained control and fail to strictly adhere to bounding box constraints. To address these limitations, we propose RSGen, a plug-and-play framework that leverages diverse edge guidance to enhance layout-driven RS image generation. Specifically, RSGen employs a progressive enhancement strategy: 1) it first enriches the diversity of edge maps composited from retrieved training instances via Image-to-Image generation; and 2) subsequently utilizes these diverse edge maps as conditioning for existing L2I models to enforce pixel-level control within bounding boxes, ensuring the generated instances strictly adhere to the layout. Extensive experiments across three baseline models demonstrate that RSGen significantly boosts the capabilities of existing L2I models. For instance, with CC-Diff on the DOTA dataset for oriented object detection, we achieve remarkable gains of +9.8/+12.0 in YOLOScore mAP50/mAP50-95 and +1.6 in mAP on the downstream detection task. Our code will be publicly available: https://github.com/D-Robotics-AI-Lab/RSGen
