Auto-Regressive (AR) models have achieved impressive results in 2D image generation by modeling joint distributions in the grid space. While this approach has been extended to the 3D domain for powerful shape generation, it still has two limitations: expensive computations on volumetric grids and ambiguous auto-regressive order along grid dimensions. To overcome these limitations, we propose the Improved Auto-regressive Model (ImAM) for 3D shape generation, which applies discrete representation learning based on a latent vector instead of volumetric grids. Our approach not only reduces computational costs but also preserves essential geometric details by learning the joint distribution in a more tractable order. Moreover, thanks to the simplicity of our model architecture, we can naturally extend it from unconditional to conditional generation by concatenating various conditioning inputs, such as point clouds, categories, images, and texts. Extensive experiments demonstrate that ImAM can synthesize diverse and faithful shapes of multiple categories, achieving state-of-the-art performance.
Unsupervised Primitive Discovery for Improved 3D Generative Modeling
Inception Institute of Artificial Intelligence(起点人工智能研究院);National University of Defense Technology(国防科技大学);Australian National University(澳大利亚国立大学);Sun Yat-sen University(中山大学);University of Canberra(堪培拉大学);Data61, Commonwealth Scientific and Industrial Research Organization(联邦科学与工业研究组织Data61)
Medical Visual Prompting (MVP): A Unified Framework for Versatile and High-Quality Medical Image Segmentation
School of Automation, Guangdong University of Technology(广东工业大学自动化学院);School of Computer Science and Technology, Guangdong University of Technology(广东工业大学计算机科学与技术学院);School of Information Science and Technology, Guangdong University of Foreign Studies(广东外语外贸大学信息科学与技术学院);Faculty of Science and Technology, University of Macau(澳门大学科技学院);The First Affiliated Hospital of Sun Yat-Sen University(中山大学附属第一医院);Sun Yat-sen University Cancer Center(中山大学肿瘤防治中心);South China Hospital, Shenzhen University(深圳大学华南医院)
MASH: Masked Anchored SpHerical Distances for 3D Shape Representation and Generation
University of Science and Technology of China(中国科学技术大学);State Key Laboratory of CAD & CG, Zhejiang University(浙江大学CAD与CG国家重点实验室);Reichman University(雷赫利亚大学);Simon Fraser University(西蒙弗雷泽大学);Shenzhen University(深圳大学)
A versatile foundation model for cine cardiac magnetic resonance image analysis tasks
UCL Hawkes Institute, Department of Medical Physics and Biomedical Engineering, University College London(伦敦大学霍克斯研究所、医学物理与生物医学工程系、伦敦大学学院);InstaDeep;Department of Computing, Imperial College London(伦敦帝国学院计算机系);Department of Brain Sciences, Imperial College London(伦敦帝国学院脑科学系);Institute of Cardiovascular Sciences, University College London(伦敦大学学院心血管科学研究所);Barts Heart Centre, Barts Health NHS Trust(巴特斯心脏中心、巴特斯健康 NHS信托)