3D-DEEP: 3-Dimensional Deep-learning based on elevation patterns for road scene interpretation

A. Hernandez; S. Woo; H. Corrales; I. Parra; E. Kim; D. F. Llorca; M. A. Sotelo

doi:10.1109/IV47402.2020.9304601

3D-DEEP: 3-Dimensional Deep-learning based on elevation patterns for road scene interpretation

A. Hernandez, S. Woo, H. Corrales, I. Parra, E. Kim, D. F. Llorca, M. A. Sotelo

Department of Electrical and Electronic Engineering

Research output: Contribution to conference › Paper › peer-review

2 Citations (Scopus)

Abstract

Road detection and segmentation is a crucial task in computer vision for safe autonomous driving. With this in mind a new net architecture (3D-DEEP) and its end-to-end training methodology for CNN-based semantic segmentation is described along this paper for. The method relies on disparity filtered and LiDAR projected images for three-dimensional information and image feature extraction through fully convolutional networks architectures. The developed models were trained and validated over Cityscapes dataset using just fine annotation examples with 19 different training classes, and over KITTI road dataset. 72.32% mean intersection over union (mIoU) has been obtained for the 19 Cityscapes training classes using the validation images. On the other hand, over KITTI dataset the model has achieved an F1 error value of 97.85% in validation and 96.02% using the test images.

Original language	English
Pages	892-898
Number of pages	7
DOIs	https://doi.org/10.1109/IV47402.2020.9304601
Publication status	Published - 2020
Event	31st IEEE Intelligent Vehicles Symposium, IV 2020 - Virtual, Las Vegas, United States Duration: 2020 Oct 19 → 2020 Nov 13

Conference

Conference	31st IEEE Intelligent Vehicles Symposium, IV 2020
Country/Territory	United States
City	Virtual, Las Vegas
Period	20/10/19 → 20/11/13

Bibliographical note

Funding Information:
This work was supported in part by Spanish Ministry of Science, Innovation and Universities (Research Grant DPI2017-90035-R) and in part by the Community Region of Madrid (Research Grant 2018/EMT-4362 SEGVAUTO 4.0-CM) and BRAVE Project, H2020, Contract #723021. This project has received funding from the Electronic Component Systems for European Leadership Joint Undertaking under grant agreement No 737469 (AutoDrive Project) and in part by the Spanish Ministry of Economy (Research Grant PCIN-2017-086). This Joint Undertaking receives support from the European Unions Horizon 2020 research and innovation programme and Germany, Austria, Spain, Italy, Latvia, Belgium, Netherlands, Sweden, Finland, Lithuania, Czech Republic, Romania, Norway. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPUs used for this Research.

Publisher Copyright:
© 2020 IEEE.

All Science Journal Classification (ASJC) codes

Computer Science Applications
Automotive Engineering
Modelling and Simulation

Access to Document

10.1109/IV47402.2020.9304601

Cite this

@conference{4878989744d9439dbcf96594aca1b13a,

title = "3D-DEEP: 3-Dimensional Deep-learning based on elevation patterns for road scene interpretation",

abstract = "Road detection and segmentation is a crucial task in computer vision for safe autonomous driving. With this in mind a new net architecture (3D-DEEP) and its end-to-end training methodology for CNN-based semantic segmentation is described along this paper for. The method relies on disparity filtered and LiDAR projected images for three-dimensional information and image feature extraction through fully convolutional networks architectures. The developed models were trained and validated over Cityscapes dataset using just fine annotation examples with 19 different training classes, and over KITTI road dataset. 72.32% mean intersection over union (mIoU) has been obtained for the 19 Cityscapes training classes using the validation images. On the other hand, over KITTI dataset the model has achieved an F1 error value of 97.85% in validation and 96.02% using the test images.",

author = "A. Hernandez and S. Woo and H. Corrales and I. Parra and E. Kim and Llorca, {D. F.} and Sotelo, {M. A.}",

note = "Funding Information: This work was supported in part by Spanish Ministry of Science, Innovation and Universities (Research Grant DPI2017-90035-R) and in part by the Community Region of Madrid (Research Grant 2018/EMT-4362 SEGVAUTO 4.0-CM) and BRAVE Project, H2020, Contract #723021. This project has received funding from the Electronic Component Systems for European Leadership Joint Undertaking under grant agreement No 737469 (AutoDrive Project) and in part by the Spanish Ministry of Economy (Research Grant PCIN-2017-086). This Joint Undertaking receives support from the European Unions Horizon 2020 research and innovation programme and Germany, Austria, Spain, Italy, Latvia, Belgium, Netherlands, Sweden, Finland, Lithuania, Czech Republic, Romania, Norway. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPUs used for this Research. Publisher Copyright: {\textcopyright} 2020 IEEE.; 31st IEEE Intelligent Vehicles Symposium, IV 2020 ; Conference date: 19-10-2020 Through 13-11-2020",

year = "2020",

doi = "10.1109/IV47402.2020.9304601",

language = "English",

pages = "892--898",

}

TY - CONF

T1 - 3D-DEEP

T2 - 31st IEEE Intelligent Vehicles Symposium, IV 2020

AU - Hernandez, A.

AU - Woo, S.

AU - Corrales, H.

AU - Parra, I.

AU - Kim, E.

AU - Llorca, D. F.

AU - Sotelo, M. A.

N1 - Funding Information: This work was supported in part by Spanish Ministry of Science, Innovation and Universities (Research Grant DPI2017-90035-R) and in part by the Community Region of Madrid (Research Grant 2018/EMT-4362 SEGVAUTO 4.0-CM) and BRAVE Project, H2020, Contract #723021. This project has received funding from the Electronic Component Systems for European Leadership Joint Undertaking under grant agreement No 737469 (AutoDrive Project) and in part by the Spanish Ministry of Economy (Research Grant PCIN-2017-086). This Joint Undertaking receives support from the European Unions Horizon 2020 research and innovation programme and Germany, Austria, Spain, Italy, Latvia, Belgium, Netherlands, Sweden, Finland, Lithuania, Czech Republic, Romania, Norway. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPUs used for this Research. Publisher Copyright: © 2020 IEEE.

PY - 2020

Y1 - 2020

N2 - Road detection and segmentation is a crucial task in computer vision for safe autonomous driving. With this in mind a new net architecture (3D-DEEP) and its end-to-end training methodology for CNN-based semantic segmentation is described along this paper for. The method relies on disparity filtered and LiDAR projected images for three-dimensional information and image feature extraction through fully convolutional networks architectures. The developed models were trained and validated over Cityscapes dataset using just fine annotation examples with 19 different training classes, and over KITTI road dataset. 72.32% mean intersection over union (mIoU) has been obtained for the 19 Cityscapes training classes using the validation images. On the other hand, over KITTI dataset the model has achieved an F1 error value of 97.85% in validation and 96.02% using the test images.

AB - Road detection and segmentation is a crucial task in computer vision for safe autonomous driving. With this in mind a new net architecture (3D-DEEP) and its end-to-end training methodology for CNN-based semantic segmentation is described along this paper for. The method relies on disparity filtered and LiDAR projected images for three-dimensional information and image feature extraction through fully convolutional networks architectures. The developed models were trained and validated over Cityscapes dataset using just fine annotation examples with 19 different training classes, and over KITTI road dataset. 72.32% mean intersection over union (mIoU) has been obtained for the 19 Cityscapes training classes using the validation images. On the other hand, over KITTI dataset the model has achieved an F1 error value of 97.85% in validation and 96.02% using the test images.

UR - http://www.scopus.com/inward/record.url?scp=85099882590&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85099882590&partnerID=8YFLogxK

U2 - 10.1109/IV47402.2020.9304601

DO - 10.1109/IV47402.2020.9304601

M3 - Paper

AN - SCOPUS:85099882590

SP - 892

EP - 898

Y2 - 19 October 2020 through 13 November 2020

ER -

3D-DEEP: 3-Dimensional Deep-learning based on elevation patterns for road scene interpretation

Abstract

Conference

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this