In the past 15 years impressive advances have been made in capturing, modeling and tracking the human body. Animals have received much less attention, despite many applications in biomechanics, biology, neuroscience, robotics, and entertainment. The main reason for the lack of 3D animalsd models is that the experience in modeling the human body cannot be easily applied to animals: animals are not collaborative, and it is not possible to bring thousands of them in a lab for 3D scanning. In this project we develop methods to learn 3D articulated statistical shape models that can represent a wide variety of species in the animal kingdom, allowing intra- and inter-species analysis of 3D shape and the automatic and non-invasive assessment of animal shape from images.
Challenges for building the model comprise defining shape correspondences and integrating data captured with different modalities, including images and video. We obtain the SMAL (Skinned Multi Animal Linear) model, a 3D articulated statistical shape model able to represent animal shapes for different species: big cats, dogs, cows, horses, zebras, and hippos. When the animal is not present in the SMAL shape space, we can still capture its 3D shape and pose with the SMALR method (SMAL with Refinement). With SMALR we can accurately capture a 3D textured mesh using a small set of uncalibrated, non-simultaneous images of the animal. The model can be fitted to images to estimate animal shape and pose.
Today animal motion is mostly captured indoors for domestic species with marker-based systems. To address this we are exploiting the 3D articulated shape model to develop a markerless motion capture system that does not require an a-priori 3D model of the subject, allowing capturing articulated motion of wild animals in their natural environment.