An Improved Thermogram Dataset for Obesity Detection Research

Obesity is a global health concern associated with severe complications, including heart disease, diabetes, and cancer. Its prevalence has seen a consistent rise, demanding early detection and intervention. The thermal imaging technique has recently been particularly promising in monitoring and early detection of obesity-related diseases. To enable research on obesity detection technology using thermal imaging, a high-quality dataset is essential. Currently, comprehensive thermogram datasets for obesity research are limited. Hence, creating an obesity thermogram dataset is critical to advancing our understanding of obesity and its treatment. In this article, we improved our previous dataset and present an open-access obesity thermogram dataset that supports the development of obesity detection systems. We describe the thermal image acquisition procedures, participant selection, room protocol configuration, and thermal camera setups. These thermal images undergo a pre-processing phase involving color mapping configuration and temperature scaling, ensuring their comparability and usability. The obesity thermogram dataset is made available for non-commercial research, fostering transparency and collaboration in the scientific community. Researchers and medical practitioners can access the dataset to train and test artificial intelligence-based obesity recognition algorithms, contributing to developing advanced medical technologies.