%0 Conference Paper %1 2019-gordon %A Gordon, Ariel %A Li, Hanhan %A Jonschkowski, Rico %A Angelova, Anelia %B 2019 IEEE/CVF International Conference on Computer Vision (ICCV) %D 2019 %K camera depth estimation monocular prediction unknown wild %P 8976-8985 %R 10.1109/ICCV.2019.00907 %T Depth From Videos in the Wild: Unsupervised Monocular Depth Learning From Unknown Cameras %U https://ieeexplore.ieee.org/document/9010905/ %X We present a novel method for simultaneous learning of depth, egomotion, object motion, and camera intrinsics from monocular videos, using only consistency across neighboring video frames as supervision signal. Similarly to prior work, our method learns by applying differentiable warping to frames and comparing the result to adjacent ones, but it provides several improvements: We address occlusions geometrically and differentiably, directly using the depth maps as predicted during training. We introduce randomized layer normalization, a novel powerful regularizer, and we account for object motion relative to the scene. To the best of our knowledge, our work is the first to learn the camera intrinsic parameters, including lens distortion, from video in an unsupervised manner, thereby allowing us to extract accurate depth and motion from arbitrary videos of unknown origin at scale. We evaluate our results on the Cityscapes, KITTI and EuRoC datasets, establishing new state of the art on depth prediction and odometry, and demonstrate qualitatively that depth prediction can be learned from a collection of YouTube videos. The code will be open sourced once anonymity is lifted.

@inproceedings{2019-gordon, abstract = {We present a novel method for simultaneous learning of depth, egomotion, object motion, and camera intrinsics from monocular videos, using only consistency across neighboring video frames as supervision signal. Similarly to prior work, our method learns by applying differentiable warping to frames and comparing the result to adjacent ones, but it provides several improvements: We address occlusions geometrically and differentiably, directly using the depth maps as predicted during training. We introduce randomized layer normalization, a novel powerful regularizer, and we account for object motion relative to the scene. To the best of our knowledge, our work is the first to learn the camera intrinsic parameters, including lens distortion, from video in an unsupervised manner, thereby allowing us to extract accurate depth and motion from arbitrary videos of unknown origin at scale. We evaluate our results on the Cityscapes, KITTI and EuRoC datasets, establishing new state of the art on depth prediction and odometry, and demonstrate qualitatively that depth prediction can be learned from a collection of YouTube videos. The code will be open sourced once anonymity is lifted.}, added-at = {2021-07-05T23:31:38.000+0200}, author = {Gordon, Ariel and Li, Hanhan and Jonschkowski, Rico and Angelova, Anelia}, biburl = {https://www.bibsonomy.org/bibtex/2ade0f55e711f59efe8ff03d02c05666d/pkoch}, booktitle = {2019 IEEE/CVF International Conference on Computer Vision (ICCV)}, doi = {10.1109/ICCV.2019.00907}, interhash = {d3656f5ab4e531819eb00f6df2f20e49}, intrahash = {ade0f55e711f59efe8ff03d02c05666d}, issn = {2380-7504}, keywords = {camera depth estimation monocular prediction unknown wild}, month = oct, pages = {8976-8985}, timestamp = {2021-07-05T23:31:38.000+0200}, title = {Depth From Videos in the Wild: Unsupervised Monocular Depth Learning From Unknown Cameras}, url = {https://ieeexplore.ieee.org/document/9010905/}, year = 2019 }