Matching-space Stereo Networks for Cross-domain Generalization
Matching-space Stereo Networks for Cross-domain Generalization
复制标题
DOI：
10.1109/3dv50981.2020.00046
复制
发表时间：
2020-10
期刊：
影响因子：
--
通讯作者：
Changjiang Cai;Matteo Poggi;S. Mattoccia;Philippos Mordohai
中科院分区：
文献类型：
--
作者：
Changjiang Cai;Matteo Poggi;S. Mattoccia;Philippos Mordohai
端到端深度网络代表了立体匹配的最新技术水平。虽然在类似于训练集的图像框架环境中表现出色，但准确性的主要下降发生在看不见的领域（例如，当从合成场景移动到真实的场景时）。在本文中，我们介绍了一种新的家庭的架构，即匹配空间网络（MS网），具有改进的泛化性能。通过将基于学习的图像RGB值特征提取替换为传统知识中的匹配函数和置信度度量，我们将学习过程从颜色空间移动到匹配空间，避免了对特定领域特征的过度专业化。在四个真实的数据集上进行的大量实验结果表明，我们的建议比传统的深度架构更能对看不见的环境进行上级推广，从而保持源域的准确性几乎不变。我们的代码可在https://qithub.com/ccj5351/MS-Nets上获得。
End-to-end deep networks represent the state of the art for stereo matching. While excelling on images framing environments similar to the training set, major drops in accuracy occur in unseen domains (e.g., when moving from synthetic to real scenes). In this paper we introduce a novel family of architectures, namely Matching-Space Networks (MS-Nets), with improved generalization properties. By replacing learning-based feature extraction from image RGB values with matching functions and confidence measures from conventional wisdom, we move the learning process from the color space to the Matching Space, avoiding over-specialization to domain specific features. Extensive experimental results on four real datasets highlight that our proposal leads to superior generalization to unseen environments over conventional deep architectures, keeping accuracy on the source domain almost unaltered. Our code is available at https://qithub.com/ccj5351/MS-Nets.