Scene Graph-Guided Generative AI Framework for Synthesizing and Evaluating Industrial Hazard Scenarios
This addresses the difficulty of obtaining accident-triggering scenario images for industrial safety applications, representing a novel method for generating synthetic hazard data.
This study tackled the problem of acquiring realistic images for training vision models to detect workplace hazards by developing a scene graph-guided generative AI framework that synthesizes photorealistic images from OSHA accident reports, with evaluation showing the proposed VQA Graph Score outperforms CLIP and BLIP metrics in discriminative sensitivity.
Training vision models to detect workplace hazards accurately requires realistic images of unsafe conditions that could lead to accidents. However, acquiring such datasets is difficult because capturing accident-triggering scenarios as they occur is nearly impossible. To overcome this limitation, this study presents a novel scene graph-guided generative AI framework that synthesizes photorealistic images of hazardous scenarios grounded in historical Occupational Safety and Health Administration (OSHA) accident reports. OSHA narratives are analyzed using GPT-4o to extract structured hazard reasoning, which is converted into object-level scene graphs capturing spatial and contextual relationships essential for understanding risk. These graphs guide a text-to-image diffusion model to generate compositionally accurate hazard scenes. To evaluate the realism and semantic fidelity of the generated data, a visual question answering (VQA) framework is introduced. Across four state-of-the-art generative models, the proposed VQA Graph Score outperforms CLIP and BLIP metrics based on entropy-based validation, confirming its higher discriminative sensitivity.