Object Shape

Object shape research focuses on accurately representing and reconstructing 3D shapes from various data sources, including images, point clouds, and sensor readings, with applications ranging from robotics to medical imaging. Current research emphasizes developing robust methods for shape estimation in challenging scenarios like occlusion and deformation, often employing deep learning architectures such as GANs, neural networks (including UNets and transformers), and diffusion models, alongside techniques like Procrustes analysis and spectral methods. These advancements improve the accuracy and efficiency of 3D shape modeling, impacting fields like automated crop monitoring, human pose estimation, and the design of physically realistic virtual and robotic systems.