XinRan-Tang/Optical-SAR-Infrared

MMDiff: Multi-modal Remote Sensing Image Generation via Cross-Modality Spatial Feature Transfer

ISPRS 2026 🔥

Haojun Tang¹ · Wenda Zhao^1,* · Hengshuai Cui¹ · Haipeng Wang²

¹ Dalian University of Technology
² Unit 92728 of PLA

^* Corresponding author:

Abstract

Collecting spatially consistent multi-modal remote sensing (MMRS) images remains challenging due to different sensors vary in the imaging principles and acquisition times. This hinders the development of data-driven MMRS technologies, which rely on large-scale training samples. This paper proposes MMDiff, the first text-driven diffusion framework explicitly designed for jointly generating structurally consistent optical (OPT), synthetic aperture radar (SAR), and infrared (IR) remote sensing images from a single text prompt via cross-modality spatial feature transfer. MMDiff first trains the OPT branch with paired optical image-text data to capture rich semantic content, and then trains the SAR/IR branches with simple modality-specific text templates to learn the corresponding style attributes, without relying on complex linguistic descriptions. Specifically, we introduce a LoRA-based modality translation adaptation mechanism to translate the style attributes of optical spatial representations to SAR and IR style attributes while preserving the underlying semantic content. The translated representations are then transferred into the SAR and IR generation branches through the proposed spatial feature transfer mechanism, enabling rich spatial details in the generated SAR/IR images while maintaining cross-modal spatial consistency. Extensive experiments demonstrate that MMDiff achieves superior image quality in terms of modality similarity and semantic consistency compared to the state-of-the-art methods. Furthermore, MMDiff benefits downstream data-driven MMRS applications, e.g., multi-modal image fusion and object classification.

From top to bottom: optical (OPT), synthetic aperture radar (SAR), and infrared (IR) images.

Dataset Overview

Optical-SAR-Infrared is a multi-modal remote sensing image-text dataset for training and evaluating text-driven generation and cross-modality spatial consistency. It provides optical (OPT), synthetic aperture radar (SAR), and infrared (IR) images with modality-specific text descriptions.

Dataset Statistics

Modality	Training pairs	Test pairs	Annotation file
Optical (OPT)	5,040	1,260	opt_train.jsonl, opt_test.jsonl
Synthetic aperture radar (SAR)	3,894	974	sar_train.jsonl, sar_test.jsonl
Infrared (IR)	2,939	738	ir_train.jsonl, ir_test.jsonl

The dataset occupies approximately 587 MB. Images are stored as JPG files and organized by modality, split, and scene category.

Directory Structure

Optical-SAR-Infrared/
├── opt/{train,test}/<scene>/*.jpg
├── sar/{train,test}/<scene>/*.jpg
├── ir/{train,test}/<scene>/*.jpg
├── opt_train.jsonl        # optical image-text annotations
├── opt_test.jsonl
├── sar_train.jsonl        # SAR image-text annotations
├── sar_test.jsonl
├── ir_train.jsonl         # infrared image-text annotations
└── ir_test.jsonl

Each JSONL record contains a relative image path and its text description:

{"file_name": "opt/train/beach/001.jpg", "text": "The beach with brown sand and the white waves washed ashore ."}

Loading the Annotations

import json
from pathlib import Path

root = Path("/path/to/Optical-SAR-Infrared")
with (root / "opt_train.jsonl").open() as f:
    samples = [json.loads(line) for line in f]

image_path = root / samples[0]["file_name"]
caption = samples[0]["text"]

Intended Uses

The dataset can be used for modality-specific diffusion training, text-to-image generation, cross-modal spatial consistency evaluation, multi-modal image fusion, and downstream remote sensing recognition tasks. Please keep the original modality and split definitions when comparing results.