MASSTAR
收藏资源简介:
MASSTAR是由中山大学、香港科技大学等机构联合提出的一个面向表面预测和补全任务的多模态大规模场景数据集。该数据集包含了超过1000个场景级别的3D网格模型,其中部分模型源自真实世界,除了3D模型外,还包含与之对应的图像、描述性文本以及点云数据。为了创建该数据集,研究人员构建了一个多功能工具链,能够高效地从复杂场景中提取高质量的3D网格模型,并生成相应的多模态数据。MASSTAR的应用领域广泛,尤其在机器人应用、高质量3D重建和自动驾驶等领域,为解决复杂场景下的表面补全问题提供了强有力的数据支撑。
MASSTAR is a multimodal large-scale scene dataset dedicated to surface prediction and completion tasks, jointly proposed by Sun Yat-sen University, The Hong Kong University of Science and Technology and other institutions. This dataset contains over 1000 scene-level 3D mesh models, some of which are sourced from real-world environments. In addition to the 3D models, it also includes corresponding images, descriptive texts and point cloud data. To develop this dataset, researchers constructed a versatile toolchain that can efficiently extract high-quality 3D mesh models from complex scenes and generate corresponding multimodal data. MASSTAR has a wide range of application scenarios, especially in robotic applications, high-quality 3D reconstruction and autonomous driving, providing robust data support for solving the surface completion problem in complex scenes.




