%0 Conference Paper %1 030-ma %A Ma, Fangchang %A Cavalheiro, Guilherme Venturelli %A Karaman, Sertac %B 2019 International Conference on Robotics and Automation (ICRA) %D 2019 %K camera completion depth lidar self-supervised %P 3288-3295 %R 10.1109/ICRA.2019.8793637 %T Self-Supervised Sparse-to-Dense: Self-Supervised Depth Completion from LiDAR and Monocular Camera %U https://ieeexplore.ieee.org/document/8793637/ %X Depth completion, the technique of estimating a dense depth image from sparse depth measurements, has a variety of applications in robotics and autonomous driving. However, depth completion faces 3 main challenges: the irregularly spaced pattern in the sparse depth input, the difficulty in handling multiple sensor modalities (when color images are available), as well as the lack of dense, pixel-level ground truth depth labels for training. In this work, we address all these challenges. Specifically, we develop a deep regression model to learn a direct mapping from sparse depth (and color images) input to dense depth prediction. We also propose a self-supervised training framework that requires only sequences of color and sparse depth images, without the need for dense depth labels. Our experiments demonstrate that the self-supervised framework outperforms a number of existing solutions trained with semi-dense annotations. Furthermore, when trained with semi-dense annotations, our network attains state-of-the-art accuracy and is the winning approach on the KITTI depth completion benchmark at the time of submission.

@inproceedings{030-ma, abstract = {Depth completion, the technique of estimating a dense depth image from sparse depth measurements, has a variety of applications in robotics and autonomous driving. However, depth completion faces 3 main challenges: the irregularly spaced pattern in the sparse depth input, the difficulty in handling multiple sensor modalities (when color images are available), as well as the lack of dense, pixel-level ground truth depth labels for training. In this work, we address all these challenges. Specifically, we develop a deep regression model to learn a direct mapping from sparse depth (and color images) input to dense depth prediction. We also propose a self-supervised training framework that requires only sequences of color and sparse depth images, without the need for dense depth labels. Our experiments demonstrate that the self-supervised framework outperforms a number of existing solutions trained with semi-dense annotations. Furthermore, when trained with semi-dense annotations, our network attains state-of-the-art accuracy and is the winning approach on the KITTI depth completion benchmark at the time of submission.}, added-at = {2021-04-12T10:04:43.000+0200}, author = {Ma, Fangchang and Cavalheiro, Guilherme Venturelli and Karaman, Sertac}, biburl = {https://www.bibsonomy.org/bibtex/2ff079407be779aa016dffab1113867fc/pkoch}, booktitle = {2019 International Conference on Robotics and Automation (ICRA)}, doi = {10.1109/ICRA.2019.8793637}, interhash = {f866756f9b4b91d57cf78c22f95d105d}, intrahash = {ff079407be779aa016dffab1113867fc}, issn = {2577-087X}, keywords = {camera completion depth lidar self-supervised}, month = may, pages = {3288-3295}, timestamp = {2021-04-12T10:04:43.000+0200}, title = {Self-Supervised Sparse-to-Dense: Self-Supervised Depth Completion from LiDAR and Monocular Camera}, url = {https://ieeexplore.ieee.org/document/8793637/}, year = 2019 }