In this paper, we address the problem of reconstructing an object's surface
from a single image using generative networks. First, we represent a 3D surface
with an aggregation of dense point clouds from multiple
使用深度神经网络从单一图像中重建 3D 点云坐标,设计了面对真实世界几何转换不变性和地面真实性模糊的问题的新型方法,包括条件形状采样器,能够预测多个可能的 3D 点云。在实验中表现优异,不仅在单图像 based 3D 重建基准测试中胜过现有技术,也在形状补全方面表现出强大性能,有望在多个可能性预测方面表现出色。