Generating Realistic X-ray Scattering Images Using Stable Diffusion and Human-in-the-loop Annotations

TL;DR

This paper presents a method to generate realistic X-ray scattering images using a fine-tuned stable diffusion model, combined with human-in-the-loop annotations to improve image quality and reduce artifacts.

Contribution

The study introduces a novel approach integrating diffusion models with human-in-the-loop corrections for scientific image generation.

Findings

High-fidelity X-ray scattering images can be generated with reduced artifacts.

Iterative training with human-reviewed images improves model accuracy.

The approach supports data augmentation and digital twin development in scientific research.

Abstract

We fine-tuned a foundational stable diffusion model using X-ray scattering images and their corresponding descriptions to generate new scientific images from given prompts. However, some of the generated images exhibit significant unrealistic artifacts, commonly known as "hallucinations". To address this issue, we trained various computer vision models on a dataset composed of 60% human-approved generated images and 40% experimental images to detect unrealistic images. The classified images were then reviewed and corrected by human experts, and subsequently used to further refine the classifiers in next rounds of training and inference. Our evaluations demonstrate the feasibility of generating high-fidelity, domain-specific images using a fine-tuned diffusion model. We anticipate that generative AI will play a crucial role in enhancing data augmentation and driving the development of digital twins in scientific research facilities.