Grounding Synthetic Data Generation With Vision and Language Models

TL;DR

This work introduces ARAS400k: A large-scale Remote sensing dataset Augmented with Synthetic data for segmentation and captioning, containing 100k real images and 300k synthetic images, each paired with segmentation maps and descriptions, and enables the automated evaluation of synthetic data.

Abstract

Deep learning models benefit from increasing data diversity and volume, motivating synthetic data augmentation to improve existing datasets. However, existing evaluation metrics for synthetic data typically calculate latent feature similarity, which is difficult to interpret and does not always correlate with the contribution to downstream tasks. We propose a vision-language grounded framework for interpretable synthetic data augmentation and evaluation in remote sensing. Our approach combines generative models, semantic segmentation and image captioning with vision and language models. Based on this framework, we introduce ARAS400k: A large-scale Remote sensing dataset Augmented with Synthetic data for segmentation and captioning, containing 100k real images and 300k synthetic images, each paired with segmentation maps and descriptions. ARAS400k enables the automated evaluation of synthetic data by analyzing semantic composition, minimizing caption redundancy, and verifying cross-modal consistency between visual structures and language descriptions. Experimental results indicate that while models trained exclusively on synthetic data reach competitive performance levels, those trained with augmented data (a combination of real and synthetic images) consistently outperform real-data baselines. Consequently, this work establishes a scalable benchmark for remote sensing tasks, specifically in semantic segmentation and image captioning. The dataset is available at zenodo.org/records/18890661 and the code base at github.com/caglarmert/ARAS400k.

Figures (5)

• Figure 1: ARAS400k synthetic data augmentation framework.
• Figure 2: Data generation pipeline, comprised of data acquisition, alignment, cleaning, pre-processing, segmentation and generative model training, caption generation and evaluation.
• Figure 3: Real and synthetic (hatched) per-class composition statistics, indicating strong similarity between real and synthetic data.
• Figure 4: t-SNE (left) and UMAP (right) comparison of real and synthetic samples, showing similarity of real and synthetic data.
• Figure 5: Qualitative comparison of training images, masks, conditioned and unconditioned synthetic samples.