OU Hudson College of Public Health Faculty Advances AI Research for Fireground Safety

Dr. Changjie Cai, a faculty member in the Department of Occupational and Environmental Health at the Hudson College of Public Health, is corresponding author of a new publication in the Journal of Occupational and Environmental Hygiene titled “Improving AI Object Detection in Fire Scenes Through Data Augmentation.”  This study is a collaborative effort among The University of Oklahoma Health Sciences Center, Embry-Riddle Aeronautical University Worldwide in the United States, and National Chengchi University in Taiwan. It explores how artificial intelligence (AI) can be enhanced to better support firefighting and disaster response operations.

As AI tools are increasingly used to support emergency responders—such as using connected cameras to identify firefighters and emergency vehicles—one major hurdle remains: poor image quality in fireground environments. Smoke, low light, and visual distortion can hinder the ability of AI to accurately detect critical elements on the scene.

The research team evaluated how two image enhancement techniques—Contrast-Limited Adaptive Histogram Equalization (CLAHE) and Zero-reference Deep Curve Estimation (Zero-DCE)—impact the accuracy of AI object detection. Their findings revealed that enhancing the visual quality of training images significantly improved detection performance.

Notably, after augmenting the training data with image enhancement techniques, the AI detector can accurately identify firefighters with a precision of 0.827 and firetrucks with a precision of 0.945. The AI model trained with CLAHE-enhanced images saw an 8% improvement in mean average precision (mAP) and a 7% increase in recall. The integration of Zero-DCE demonstrated particular efficacy in recognizing firetrucks in low-light conditions, achieving the highest precision value of 0.945 among all the cases considered.

This research highlights how improved training datasets, combined with image enhancement, can help AI models more reliably identify key objects like firefighters and emergency vehicles during real-time fireground operations. The authors also propose future directions for researchers working to optimize AI for disaster response.

This work not only demonstrates the innovative contributions of Dr. Cai and his colleagues to public health and safety but also underscores the growing role of AI in supporting first responders in dangerous and dynamic environments.

You can read the full article at https://www.tandfonline.com/doi/full/10.1080/15459624.2025.2499600