DepthSynth：一种2.5D识别中基于CAD模型的真实感数据实时生成框架

每天一分钟，带你读遍机器人顶级会议文章

标题：DepthSynth: Real-Time Realistic Synthetic Data Generation from CAD Models

for 2.5D Recognition

作者：Benjamin Planche, Ziyan Wu, Kai Ma, Shanhui Sun，Stefan Kluckner，Oliver Lehmann

来源：3dv 2017 (International Conference on 3D Vision）

播音员：郭晨

编译：颜青松

欢迎个人转发朋友圈；其他机构或自媒体如需转载，后台留言申请授权

摘要

目前，深度神经网络已经成为计算机视觉的主导，需要使用大量的标签数据。不过由于标签数据的收集是一个非常枯燥乏味的过程，且在很多情况下根本不可能收集到足够多的标签数据。为解决此问题，大量深度神经网络开始运用合成数据的方法来训练网络。

对于本文中提到的深度图而言，真实扫描数据总是存在偏差，会给网络的精度带来较大的影响。因此，本文提出了一种端到端的数据合成框架，能够完整模拟数据采集设备的采集机制，顾虑传感器噪声、材料反射率、表面几何形状等等重要因素，从三维模型中生成逼真的深度数据。

与之前的方法相比，本文中的算法模拟了更多类型的传感器，并且在经过改良的评价方法下具有更高的真实度；除此之外，本文还测量了针对不同识别任务合成数据对神经网络的影响，展示了本文中的方法能够无缝地和这些神经网络集成在一起，并且提高它们的精度。

下图展示了文中算法的整个流程，展示了如何从CAD模型逐步生成深度图的过程。

下图是使用文中的框架生成的一些合成数据。

Abstract

Recent progress in computer vision has been dominated by deep neural networks trained over large amounts of labeled data. Collecting such datasets is however a tedious, often impossible task; hence a surge in approaches relying solely on synthetic data for their training.

For depth images however, discrepancies with real scans still noticeably affect the end performance. We thus propose an end-to-end framework which simulates the whole mechanism of these devices, generating realistic depth data from 3D models by comprehensively modeling vital factors e.g. sensor noise, material reflectance, surface geometry.

Not only does our solution cover a wider range of sensors and achieve more realistic results than previous methods, assessed through extended evaluation, but we go further by measuring the impact on the training of neural networks for various recognition tasks; demonstrating how our pipeline seamlessly integrates such architectures and consistently enhances their performance.