Learning to Adapt for Stereo
Learning to Adapt for Stereo
复制标题
DOI：
10.1109/cvpr.2019.00989
复制
发表时间：
2019-04
期刊：
影响因子：
--
通讯作者：
A. Tonioni;Oscar Rahnama;Thomas Joy;L. D. Stefano;Thalaiyasingam Ajanthan;Philip H. S. Torr
中科院分区：
文献类型：
--
作者：
A. Tonioni;Oscar Rahnama;Thomas Joy;L. D. Stefano;Thalaiyasingam Ajanthan;Philip H. S. Torr
立体深度估计的真实的世界应用需要对环境中的动态变化鲁棒的模型。尽管基于深度学习的立体方法是成功的，但它们往往无法推广到环境中看不见的变化，使它们不太适合自动驾驶等实际应用。在这项工作中，我们引入了一个“学习适应”框架，使深度立体方法能够以无监督的方式不断适应新的目标域。具体而言，我们的方法将适应过程纳入学习目标，以获得更适合无监督在线适应的基本参数集。为了进一步提高自适应的质量，我们学习了一个置信度，可以有效地掩盖在无监督自适应过程中引入的错误。我们在合成和真实世界的立体数据集上评估了我们的方法，我们的实验证明，学习适应确实有利于在不同领域的在线适应。
Real world applications of stereo depth estimation require models that are robust to dynamic variations in the environment. Even though deep learning based stereo methods are successful, they often fail to generalize to unseen variations in the environment, making them less suitable for practical applications such as autonomous driving. In this work, we introduce a ``learning-to-adapt'' framework that enables deep stereo methods to continuously adapt to new target domains in an unsupervised manner. Specifically, our approach incorporates the adaptation procedure into the learning objective to obtain a base set of parameters that are better suited for unsupervised online adaptation. To further improve the quality of the adaptation, we learn a confidence measure that effectively masks the errors introduced during the unsupervised adaptation. We evaluate our method on synthetic and real-world stereo datasets and our experiments evidence that learning-to-adapt is, indeed beneficial for online adaptation on vastly different domains.