arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.13258cs.CVcs.AI
收藏

用于单目深度预测的边界诱导和场景聚合网络

Boundary-induced and scene-aggregated network for monocular depth prediction

Beijing University of Posts and Telecommunications(北京邮电大学);Huazhong University of Science and Technology(华中科技大学)

Feng Xue, Junfeng Cao, Yu Zhou, Fei Sheng, Yankai Wang, Anlong Ming

摘要:

单目深度预测是场景理解中的一项重要任务。它旨在预测单个RGB图像的密集深度。随着深度学习的发展,这个任务的性能有了很大的提升。然而,有两个问题尚未解决:(1)深度特征编码错误的场景中最远的区域,这导致预测深度的失真3D结构;(2)低级特征未得到充分利用,这使得在深度突然变化的情况下更难以估计边缘附近的深度。为了解决这两个问题,我们提出了边界诱导和场景聚合网络(BS-Net)。在该网络中,深度相关编码器(DCE)首先被设计为获得图像中的区域之间的上下文相关性,并通过考虑相关性来感知最远的区域。同时,自底向上边界融合(BUBF)模块的设计,提取准确的边界,指示深度变化。最后,设计了条纹细化模块(SRM),对边界线索诱导的密集深度进行细化,提高了预测深度的边界精度。在NYUD v2数据集和\xff{the iBims-1 dataset}上的几个实验结果说明了所提出的方法的最先进的性能。并利用SUN-RGBD数据集对该方法的推广性进行了评估。代码可在https://github.com/XuefengBUPT/BS-Net上获得。

英文摘要:

Monocular depth prediction is an important task in scene understanding. It aims to predict the dense depth of a single RGB image. With the development of deep learning, the performance of this task has made great improvements. However, two issues remain unresolved: (1) The deep feature encodes the wrong farthest region in a scene, which leads to a distorted 3D structure of the predicted depth; (2) The low-level features are insufficient utilized, which makes it even harder to estimate the depth near the edge with sudden depth change. To tackle these two issues, we propose the Boundary-induced and Scene-aggregated network (BS-Net). In this network, the Depth Correlation Encoder (DCE) is first designed to obtain the contextual correlations between the regions in an image, and perceive the farthest region by considering the correlations. Meanwhile, the Bottom-Up Boundary Fusion (BUBF) module is designed to extract accurate boundary that indicates depth change. Finally, the Stripe Refinement module (SRM) is designed to refine the dense depth induced by the boundary cue, which improves the boundary accuracy of the predicted depth. Several experimental results on the NYUD v2 dataset and \xff{the iBims-1 dataset} illustrate the state-of-the-art performance of the proposed approach. And the SUN-RGBD dataset is employed to evaluate the generalization of our method. Code is available at https://github.com/XuefengBUPT/BS-Net.

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。

↑