arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.01148cs.CVcs.LGstat.ML
收藏

基于无监督体分割的三维场景对象分解

Decomposing 3D Scenes into Objects via Unsupervised Volume Segmentation

TU Darmstadt(达姆施塔特工业大学);DeepMind

Karl Stelzner, Kristian Kersting, Adam R. Kosiorek

摘要:

我们提出了ObSuRF,一种将场景的单个图像转换为表示为一组神经辐射场(NeRF)的3D模型的方法,每个NeRF对应于不同的对象。编码器网络的单个前向传递输出描述场景中的对象的一组潜在向量。这些向量被独立地用于调节NeRF解码器,定义每个对象的几何形状和外观。我们通过推导一种新的损失来提高学习的计算效率,这种损失允许在RGB-D输入上训练NeRF,而无需显式的光线行进。在确认该模型在三个2D图像分割基准上的表现等于或优于最新技术水平后,我们将其应用于两个多对象3D数据集:CLEVR的多视图版本和一个新的数据集,其中场景由ShapeNet模型填充。我们发现,在训练场景的RGB-D视图上训练ObSuRF后,它不仅能够恢复单个输入图像中描绘的场景的3D几何形状,而且还能够将其分割成对象,尽管在这方面没有受到监督。

英文摘要:

We present ObSuRF, a method which turns a single image of a scene into a 3D model represented as a set of Neural Radiance Fields (NeRFs), with each NeRF corresponding to a different object. A single forward pass of an encoder network outputs a set of latent vectors describing the objects in the scene. These vectors are used independently to condition a NeRF decoder, defining the geometry and appearance of each object. We make learning more computationally efficient by deriving a novel loss, which allows training NeRFs on RGB-D inputs without explicit ray marching. After confirming that the model performs equal or better than state of the art on three 2D image segmentation benchmarks, we apply it to two multi-object 3D datasets: A multiview version of CLEVR, and a novel dataset in which scenes are populated by ShapeNet models. We find that after training ObSuRF on RGB-D views of training scenes, it is capable of not only recovering the 3D geometry of a scene depicted in a single input image, but also to segment it into objects, despite receiving no supervision in that regard.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑