Dynamo-Depth: Fixing Unsupervised Depth Estimation for Dynamical Scenes-猫眼课题宝

DOI：

10.48550/arxiv.2310.18887

发表时间：

2023-10

期刊：

ArXiv

影响因子：

通讯作者：

中科院分区：

其他

文献类型：

作者：

Yihong Sun;Bharath Hariharan

无监督单目深度估计技术已经证明了令人鼓舞的结果，但通常假设场景是静态的。这些技术在动态场景中训练时会受到影响，其中明显的物体运动同样可以通过假设物体的独立运动或通过改变其深度来解释。这种模糊性导致深度估计器预测移动对象的错误深度。为了解决这个问题，我们引入了Dynamo-Depth，这是一种统一的方法，通过联合学习单目深度，3D独立流场和未标记单目视频的运动分割来消除动态运动的歧义。具体来说，我们提供了我们的关键见解，即运动分割的良好初始估计足以联合学习深度和独立运动，尽管基本的潜在模糊性。我们提出的方法在Waymo Open和nuScenes数据集上实现了最先进的单目深度估计性能，并在移动对象的深度方面有了显着改善。代码和其他结果可在https://dynamo-depth.github.io上获得。

Unsupervised monocular depth estimation techniques have demonstrated encouraging results but typically assume that the scene is static. These techniques suffer when trained on dynamical scenes, where apparent object motion can equally be explained by hypothesizing the object's independent motion, or by altering its depth. This ambiguity causes depth estimators to predict erroneous depth for moving objects. To resolve this issue, we introduce Dynamo-Depth, an unifying approach that disambiguates dynamical motion by jointly learning monocular depth, 3D independent flow field, and motion segmentation from unlabeled monocular videos. Specifically, we offer our key insight that a good initial estimation of motion segmentation is sufficient for jointly learning depth and independent motion despite the fundamental underlying ambiguity. Our proposed method achieves state-of-the-art performance on monocular depth estimation on Waymo Open and nuScenes Dataset with significant improvement in the depth of moving objects. Code and additional results are available at https://dynamo-depth.github.io.

Dynamo-Depth: Fixing Unsupervised Depth Estimation for Dynamical Scenes

文献摘要

相似文献