14. Main stream
単眼深度推定の歴史
14
SfM-Net
Vijayanarasimhan
et al.
(2017)
SfM Learner
Zhou et al.
(CVPR 2017)
Monodepth
Godard et al.
(CVPR 2017)
Edge-aware
Depth-Normal
Consistency
Yang et al.
(AAAI 2018) LKVO Learner
Wang et al.
(CVPR 2018)
GeoNet
Yin, Shi
(CVPR 2018)
LEGO
Yang et al.
(CVPR 2018)
Monodepth2
Godard et al.
(2018)
Depth from
Videos in the Wild
Gordon et al.
(2019)
Every Pixel Counts
Yang et al.
(CVPR 2018)
Vid2depth
Mahjourian et al.
(CVPR 2018)
+ Optical flow
Stereo +
Spatial
Transformer
Estimate Depth
and Pose.
Improve
smoothness
loss
Improve
PoseNet
ICP
matching
loss
Estimate
Camera
Matrix (K)
※私見です
15. SfM Learner
15
書誌情報
Unsupervised Learning of Depth and Ego-Motion from Video
著者:Tinghui Zhou, Matthew Brown, Noah Snavely, David Lowe
CVPR 2017 (Oral)
https://people.eecs.berkeley.edu/~tinghuiz/projects/SfMLearner/
概要
単眼動画のみから深度推定を学習する
Depth CNN と Pose CNN を持ち,前者が深度を,後者が自己位置
の変化(カメラ外部行列)を推定する
Bilinear sampler を用いて画像を歪め,隣接フレームを合成する
32. 参考文献
32
• Godard, C., Mac Aodha, O., & Brostow, G. (2018). Digging Into Self-Supervised Monocular Depth
Estimation. arXiv:1806.01260 [cs, stat]. http://arxiv.org/abs/1806.01260. Accessed 27 April 2019
• Godard, C., Mac Aodha, O., & Brostow, G. J. (2016). Unsupervised Monocular Depth Estimation with
Left-Right Consistency. arXiv:1609.03677 [cs, stat]. http://arxiv.org/abs/1609.03677. Accessed 27 April
2019
• Gordon, A., Li, H., Jonschkowski, R., & Angelova, A. (2019). Depth from Videos in the Wild:
Unsupervised Monocular Depth Learning from Unknown Cameras. arXiv:1904.04998 [cs].
http://arxiv.org/abs/1904.04998. Accessed 27 April 2019
• Mahjourian, R., Wicke, M., & Angelova, A. (2018). Unsupervised Learning of Depth and Ego-Motion
from Monocular Video Using 3D Geometric Constraints. arXiv:1802.05522 [cs].
http://arxiv.org/abs/1802.05522. Accessed 27 April 2019
• Ummenhofer, B., Zhou, H., Uhrig, J., Mayer, N., Ilg, E., Dosovitskiy, A., & Brox, T. (2016). DeMoN: Depth
and Motion Network for Learning Monocular Stereo. 2017 IEEE Conference on Computer Vision and
Pattern Recognition (CVPR), 5622–5631. doi:10.1109/CVPR.2017.596
• Vijayanarasimhan, S., Ricco, S., Schmid, C., Sukthankar, R., & Fragkiadaki, K. (2017). SfM-Net: Learning
of Structure and Motion from Video. arXiv:1704.07804 [cs]. http://arxiv.org/abs/1704.07804. Accessed 7
May 2019
33. 参考文献
33
• Wang, C., Buenaposada, J. M., Zhu, R., & Lucey, S. (2017). Learning Depth from Monocular Videos
using Direct Methods. arXiv:1712.00175 [cs]. http://arxiv.org/abs/1712.00175. Accessed 27 April 2019
• Yang, Z., Wang, P., Wang, Y., Xu, W., & Nevatia, R. (2018a). LEGO: Learning Edge with Geometry all at
Once by Watching Videos. arXiv:1803.05648 [cs]. http://arxiv.org/abs/1803.05648. Accessed 27 April
2019
• Yang, Z., Wang, P., Wang, Y., Xu, W., & Nevatia, R. (2018b). Every Pixel Counts: Unsupervised Geometry
Learning with Holistic 3D Motion Understanding. arXiv:1806.10556 [cs]. http://arxiv.org/abs/1806.10556.
Accessed 27 April 2019
• Yang, Z., Wang, P., Xu, W., Zhao, L., & Nevatia, R. (2017). Unsupervised Learning of Geometry with
Edge-aware Depth-Normal Consistency. arXiv:1711.03665 [cs]. http://arxiv.org/abs/1711.03665.
Accessed 27 April 2019
• Yin, Z., & Shi, J. (2018). GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera
Pose. arXiv:1803.02276 [cs]. http://arxiv.org/abs/1803.02276. Accessed 27 April 2019
• Zhou, T., Brown, M., Snavely, N., & Lowe, D. G. (2017). Unsupervised Learning of Depth and Ego-
Motion from Video. arXiv:1704.07813 [cs]. http://arxiv.org/abs/1704.07813. Accessed 27 April 2019