arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.02708cs.CV
收藏

自校正网络用于室内无监督深度估计

Auto-Rectify Network for Unsupervised Indoor Depth Estimation

The University of Adelaide(阿德莱德大学);Australian Centre for Robotic Vision(澳大利亚机器人视觉中心);TuSimple(图森未来)

Jia-Wang Bian, Huangying Zhan, Naiyan Wang, Tat-Jun Chin, Chunhua Shen, Ian Reid

摘要:

使用从未标记视频训练的CNN进行单视图深度估计已经显示出巨大的前景。然而,优秀的结果大多是在街景驾驶场景中获得的,并且这样的方法在其他设置中通常失败,特别是通过手持设备拍摄的室内视频。在这项工作中,我们建立了复杂的自我运动表现在手持设置是学习深度的一个关键障碍。我们的基本分析表明,旋转在训练过程中表现为噪声,而不是提供监督信号的平移(基线)。为了解决这个问题,我们提出了一种数据预处理方法,通过去除训练图像的相对旋转来校正训练图像,以实现有效的学习。业绩的显著改善验证了我们的动机。为了实现端到端学习而无需预处理,我们提出了一种具有新颖损失函数的自动校正网络,它可以在训练过程中自动学习校正图像。因此,我们的结果在具有挑战性的NYUv 2数据集上远远优于以前的无监督SOTA方法。我们还展示了我们训练的模型在ScanNet和Make 3D中的泛化能力,以及我们提出的学习方法在7-Scenes和KITTI数据集上的通用性。

英文摘要:

Single-View depth estimation using the CNNs trained from unlabelled videos has shown significant promise. However, excellent results have mostly been obtained in street-scene driving scenarios, and such methods often fail in other settings, particularly indoor videos taken by handheld devices. In this work, we establish that the complex ego-motions exhibited in handheld settings are a critical obstacle for learning depth. Our fundamental analysis suggests that the rotation behaves as noise during training, as opposed to the translation (baseline) which provides supervision signals. To address the challenge, we propose a data pre-processing method that rectifies training images by removing their relative rotations for effective learning. The significantly improved performance validates our motivation. Towards end-to-end learning without requiring pre-processing, we propose an Auto-Rectify Network with novel loss functions, which can automatically learn to rectify images during training. Consequently, our results outperform the previous unsupervised SOTA method by a large margin on the challenging NYUv2 dataset. We also demonstrate the generalization of our trained model in ScanNet and Make3D, and the universality of our proposed learning method on 7-Scenes and KITTI datasets.

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。

↑