A data-efficient framework that turns unlabeled overhead imagery into paired editing supervision — editing the region you select while preserving the surrounding scene.
Satellite image editing requires spatially precise object-level control, but supervised editing datasets for overhead imagery are costly to build because object masks, semantic labels, and paired edits are rarely available at scale. We introduce SatEdit, a mask-conditioned satellite image editing framework that constructs training supervision from unlabeled imagery. SatEdit proposes object masks with a segmentation foundation model, assigns semantic labels to sampled segments with a Vision-Language Model, and applies lightweight human verification before generating paired addition and removal examples through mask-guided inpainting. We fine-tune a high-resolution image editing backbone with LoRA on a SODA-A-derived dataset containing 1,014 images and 852 verified object annotations across 91 classes. In controlled comparisons with open-source and proprietary image editing models, SatEdit achieves the highest aggregate masked-region semantic alignment, with a CLIP score of 0.6322 and CLIP delta of 0.0726, while preserving the surrounding scene qualitatively.
Rather than manually authoring before/after pairs, SatEdit decomposes supervision into spatial support, object semantics, label verification, and edited-image synthesis.
SAM2 generates dense, class-agnostic object masks over each satellite scene; area-based filtering and subsampling keep the informative segments.
Qwen3-VL labels each mask-aware crop from a fixed satellite-object vocabulary — constrained classification that reduces hallucination.
A single-pass check turns annotation into validation: annotators accept or correct the predicted label, not create it from scratch.
Mask-guided inpainting (LaMA) converts verified segments into paired addition and removal examples with source image, mask, and instruction.
SatEdit obtains the highest CLIP and CLIP Δ across addition, removal, and the aggregate — strongest prompt-aligned editing inside the mask. Nano Banana 2 remains the most conservative editor, changing the background least.
Real editing outputs from our benchmark. Across examples, SatEdit confines the requested change to the masked region while baselines more often spill outside the mask, place objects at implausible scale, or leave the edit incomplete. Click any tile to enlarge.
@article{talha2026satedit,
title = {SatEdit: Mask-Conditioned Image Editing via VLM-Guided Segment Annotation},
author = {Talha, Muhammad and Amer, Muhammad Ahmed},
year = {2026}
}