This project aims at accurately estimating the body shape of persons wearing casual clothing from their 3D scans. In existing works, the human body shape and pose are commonly represented by a low-dimensional mathematical model learned from a database of 3D scans. The model is then fitted to a target scan to estimate the underlying body shape. However, the fitting usually lacks accuracy as too simple assumptions are made, the data is of low quality or only a rough estimate is targeted. Moreover, the process systematically requires manual intervention. This project will produce novel methods for 3D human body shape analysis. First, new clues in the data will be exploited to more accurately constrain the possible body shape under clothing. Then, novel models will be proposed to estimate accurate biometric measurements. Furthermore, methods will be devised to automate the preparation of a dataset of full-body 3D scans of people. In parallel to that, a unique high-resolution dataset of people wearing clothing will be collected to evaluate the methods and develop new industrial applications. The results of the project will enhance the accuracy of body shape recovery, steering applications towards a safer and healthier society. The security level in airports, public places and online platforms can be increased with improved systems for person identification and suspicious object detection. In healthcare, better services and monitoring can be offered with an automated and accurate system for taking full-body biometrical measurements. In sport and fashion industries, higher-quality products can be obtained by designing tailored garments based on accurate customised measurements. Moreover, the new methods will simplify and make accessible the creation of high-quality datasets of 3D scans of people. Finally, the results of the project are expected to be transferable to other tasks in computer vision and computer graphics.