Scene Synthesis via Uncertainty-Driven Attribute Synchronization
Scene Synthesis via Uncertainty-Driven Attribute Synchronization
复制标题
DOI：
10.1109/iccv48922.2021.00558
复制
发表时间：
2021-08
期刊：
影响因子：
--
通讯作者：
Haitao Yang;Zaiwei Zhang;Siming Yan;Haibin Huang;Chongyang Ma;Yi Zheng;Chandrajit L. Bajaj;Qi-Xing Huang
中科院分区：
文献类型：
--
作者：
Haitao Yang;Zaiwei Zhang;Siming Yan;Haibin Huang;Chongyang Ma;Yi Zheng;Chandrajit L. Bajaj;Qi-Xing Huang
开发深度神经网络来生成3D场景是神经合成中的一个基本问题，可直接应用于建筑CAD，计算机图形学以及生成虚拟机器人训练环境。这项任务是具有挑战性的，因为3D场景表现出不同的模式，从连续的，如对象大小和形状对之间的相对姿态，离散的模式，如对称关系的对象的出现和共现。本文介绍了一种新的神经场景合成方法，可以捕捉不同的特征模式的三维场景。我们的方法结合了基于神经网络和传统场景合成方法的优势。我们使用从训练数据中学习的参数先验分布，它提供了对象属性和相对属性的不确定性，以正则化前馈神经模型的输出。此外，我们的方法不是仅仅预测场景布局，而是预测一组过完备的属性。这种方法允许我们利用预测属性之间的基本一致性约束来修剪不可行的预测。实验结果表明，我们的方法大大优于现有的方法。生成的3D场景忠实地插值训练数据，同时保留连续和离散的特征模式。
Developing deep neural networks to generate 3D scenes is a fundamental problem in neural synthesis with immediate applications in architectural CAD, computer graphics, as well as in generating virtual robot training environments. This task is challenging because 3D scenes exhibit diverse patterns, ranging from continuous ones, such as object sizes and the relative poses between pairs of shapes, to discrete patterns, such as occurrence and co-occurrence of objects with symmetrical relationships. This paper introduces a novel neural scene synthesis approach that can capture diverse feature patterns of 3D scenes. Our method combines the strength of both neural network-based and conventional scene synthesis approaches. We use the parametric prior distributions learned from training data, which provide uncertainties of object attributes and relative attributes, to regularize the outputs of feed-forward neural models. Moreover, instead of merely predicting a scene layout, our approach predicts an over-complete set of attributes. This methodology allows us to utilize the underlying consistency constraints among the predicted attributes to prune infeasible predictions. Experimental results show that our approach outperforms existing methods considerably. The generated 3D scenes interpolate the training data faithfully while preserving both continuous and discrete feature patterns.