Skip to content
Back to outputs

Two-stage deep regression enhanced depth estimation from a single RGB image

Research output: Contribution to journalArticlepeer-review

Standard

Two-stage deep regression enhanced depth estimation from a single RGB image. / Sun, Jianyuan; Wang, Zidong; Yu, Hui; Zhang, Shu; Dong, Junyu; Gao, Pengxiang.

In: IEEE Transactions on Emerging Topics in Computing, 28.10.2020.

Research output: Contribution to journalArticlepeer-review

Harvard

Sun, J, Wang, Z, Yu, H, Zhang, S, Dong, J & Gao, P 2020, 'Two-stage deep regression enhanced depth estimation from a single RGB image', IEEE Transactions on Emerging Topics in Computing. https://doi.org/10.1109/TETC.2020.3034559

APA

Sun, J., Wang, Z., Yu, H., Zhang, S., Dong, J., & Gao, P. (2020). Two-stage deep regression enhanced depth estimation from a single RGB image. IEEE Transactions on Emerging Topics in Computing. https://doi.org/10.1109/TETC.2020.3034559

Vancouver

Sun J, Wang Z, Yu H, Zhang S, Dong J, Gao P. Two-stage deep regression enhanced depth estimation from a single RGB image. IEEE Transactions on Emerging Topics in Computing. 2020 Oct 28. https://doi.org/10.1109/TETC.2020.3034559

Author

Sun, Jianyuan ; Wang, Zidong ; Yu, Hui ; Zhang, Shu ; Dong, Junyu ; Gao, Pengxiang. / Two-stage deep regression enhanced depth estimation from a single RGB image. In: IEEE Transactions on Emerging Topics in Computing. 2020.

Bibtex

@article{35155d43893a466e904bf42001dae2e8,
title = "Two-stage deep regression enhanced depth estimation from a single RGB image",
abstract = "Depth estimation plays a significant role in industrial applications, e.g. augmented reality, robotic mapping and autonomous driving. Traditional approaches for capturing depth, such as laser or depth sensor based methods, are difficult to use in most scenarios due to the limitations of high system cost and limited operational conditions. As an inexpensive and convenient approach, using the computational models to estimate depth from a single RGB image offers a preferable way for the depth prediction. Although the design of computational models to estimate the depth map has been widely investigated, the majority of models suffers from low prediction accuracy due to the sole utilization of a one-stage regression strategy. Inspired by both theoretical and practical success of two-stage regression, we propose a two-stage deep regression model, which is composed of two state-of-the-art network architectures, i.e. the fully convolutional residual network (FCRN) and the conditional generation adversarial network (cGAN). FCRN has been proved to possess a strong prediction ability for depth prediction, but fine details in the depth map are still incomplete. Accordingly, we have improved the existing cGAN model to refine the FCRN-based depth prediction. The experimental results show that the proposed two-stage deep regression model outperforms existing state-of-the-art methods.",
keywords = "RCUK, EPSRC, EP/N025849/1, depth prediction, a single RGB image, the rough depth map, neural networks",
author = "Jianyuan Sun and Zidong Wang and Hui Yu and Shu Zhang and Junyu Dong and Pengxiang Gao",
year = "2020",
month = oct,
day = "28",
doi = "10.1109/TETC.2020.3034559",
language = "English",
journal = "IEEE Transactions on Emerging Topics in Computing",
issn = "2168-6750",
publisher = "IEEE",

}

RIS

TY - JOUR

T1 - Two-stage deep regression enhanced depth estimation from a single RGB image

AU - Sun, Jianyuan

AU - Wang, Zidong

AU - Yu, Hui

AU - Zhang, Shu

AU - Dong, Junyu

AU - Gao, Pengxiang

PY - 2020/10/28

Y1 - 2020/10/28

N2 - Depth estimation plays a significant role in industrial applications, e.g. augmented reality, robotic mapping and autonomous driving. Traditional approaches for capturing depth, such as laser or depth sensor based methods, are difficult to use in most scenarios due to the limitations of high system cost and limited operational conditions. As an inexpensive and convenient approach, using the computational models to estimate depth from a single RGB image offers a preferable way for the depth prediction. Although the design of computational models to estimate the depth map has been widely investigated, the majority of models suffers from low prediction accuracy due to the sole utilization of a one-stage regression strategy. Inspired by both theoretical and practical success of two-stage regression, we propose a two-stage deep regression model, which is composed of two state-of-the-art network architectures, i.e. the fully convolutional residual network (FCRN) and the conditional generation adversarial network (cGAN). FCRN has been proved to possess a strong prediction ability for depth prediction, but fine details in the depth map are still incomplete. Accordingly, we have improved the existing cGAN model to refine the FCRN-based depth prediction. The experimental results show that the proposed two-stage deep regression model outperforms existing state-of-the-art methods.

AB - Depth estimation plays a significant role in industrial applications, e.g. augmented reality, robotic mapping and autonomous driving. Traditional approaches for capturing depth, such as laser or depth sensor based methods, are difficult to use in most scenarios due to the limitations of high system cost and limited operational conditions. As an inexpensive and convenient approach, using the computational models to estimate depth from a single RGB image offers a preferable way for the depth prediction. Although the design of computational models to estimate the depth map has been widely investigated, the majority of models suffers from low prediction accuracy due to the sole utilization of a one-stage regression strategy. Inspired by both theoretical and practical success of two-stage regression, we propose a two-stage deep regression model, which is composed of two state-of-the-art network architectures, i.e. the fully convolutional residual network (FCRN) and the conditional generation adversarial network (cGAN). FCRN has been proved to possess a strong prediction ability for depth prediction, but fine details in the depth map are still incomplete. Accordingly, we have improved the existing cGAN model to refine the FCRN-based depth prediction. The experimental results show that the proposed two-stage deep regression model outperforms existing state-of-the-art methods.

KW - RCUK

KW - EPSRC

KW - EP/N025849/1

KW - depth prediction

KW - a single RGB image

KW - the rough depth map

KW - neural networks

U2 - 10.1109/TETC.2020.3034559

DO - 10.1109/TETC.2020.3034559

M3 - Article

JO - IEEE Transactions on Emerging Topics in Computing

JF - IEEE Transactions on Emerging Topics in Computing

SN - 2168-6750

ER -

ID: 23092167