Reliable 3D body shape assessment is necessary for many applications such as health monitoring and virtual try-on. However, scanning the body under extremely small clothes is a major challenge in this technical research. Relevant researchers proposed to solve this technical problem by using 3D scanning methods, and proposed the BUFF dataset for quantitative evaluation.

BUFF dataset, full name of Bodies Under Flowing Fashion, is a high-quality four-dimensional dataset. A total of 3 men and 2 women participated in the data collection. They wore two styles of clothing: T-shirts and trousers or football suits, and made three different movements: hips, tilt_twist_left, shoulders_mill.