Body Shape

Body shape research focuses on accurately representing and manipulating 3D human body models from various input modalities, such as images and videos, aiming for realistic and consistent results across different poses and clothing. Current research emphasizes developing robust algorithms and model architectures, including neural networks (e.g., diffusion models, Transformers, and graph neural networks), to address challenges like occlusions, clothing variations, and diverse body types. This work has significant implications for various fields, including virtual reality, fashion technology, healthcare (e.g., liver fat estimation), and computer vision, by enabling more accurate and personalized applications.