Google develops AI to estimate body fat via smartphone photos

Google is developing an artificial intelligence system called ‘PhotoScan’ designed to estimate body fat percentage using standard smartphone photographs. The process requires images of the body from both the front and side, supplemented by the user’s height and weight.

Beyond calculating total body fat, the technology aims to identify fat distribution, such as abdominal fat, which is often a more significant indicator of health risks than the Body Mass Index (BMI) can provide. In comparative testing, the AI’s estimates were found to be more accurate than bioimpedance analysis (BIA)—the technology used in smartwatches and smart scales—averaging within 2.15 percentage points of medical reference values, compared to 2.91 percentage points for BIA.