SinBMI: Estimating BMI from a Single Image

Body mass index is a quantitative measure that relates an individual’s weight to their height, providing a standardized indicator of body composition. This metric has been extensively validated and is considered a reliable biometric tool in medical evaluations. However, traditional methods to calculate BMI require direct measurements or self-reported data, which can be time-consuming and subject to error. Therefore, there is great interest in non-apparatus and remote BMI measurement approaches. In this paper, we introduce SinBMI, a novel deep learning model that estimates human BMI from a single image. SinBMI is more than an order of magnitude lighter than existing single-image BMI estimation models, making it highly suitable for mobile deployment. SinBMI achieves state-of-the-art performance on the largest dataset to date, with an MAPE of 9.38% and an MAE of 2.92.