Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement

Minhyeok Heo, Jaehan Lee, Kyung Rae Kim, Han Ul Kim, Chang-Su Kim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We propose a monocular depth estimation algorithm based on whole strip masking (WSM) and reliability-based refinement. First, we develop a convolutional neural network (CNN) tailored for the depth estimation. Specifically, we design a novel filter, called WSM, to exploit the tendency that a scene has similar depths in horizonal or vertical directions. The proposed CNN combines WSM upsampling blocks with a ResNet encoder. Second, we measure the reliability of an estimated depth, by appending additional layers to the main CNN. Using the reliability information, we perform conditional random field (CRF) optimization to refine the estimated depth map. Experimental results demonstrate that the proposed algorithm provides the state-of-the-art depth estimation performance.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
EditorsVittorio Ferrari, Cristian Sminchisescu, Yair Weiss, Martial Hebert
PublisherSpringer Verlag
Pages39-55
Number of pages17
ISBN (Print)9783030012243
DOIs
Publication statusPublished - 2018 Jan 1
Event15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany
Duration: 2018 Sep 82018 Sep 14

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11208 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th European Conference on Computer Vision, ECCV 2018
CountryGermany
CityMunich
Period18/9/818/9/14

Fingerprint

Depth Estimation
Masking
Strip
Refinement
Neural Networks
Neural networks
Conditional Random Fields
Depth Map
Encoder
Estimation Algorithms
Vertical
Filter
Optimization
Experimental Results
Demonstrate

Keywords

  • Depth map refinement
  • Monocular depth estimation
  • Reliability
  • Whole strip masking

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Heo, M., Lee, J., Kim, K. R., Kim, H. U., & Kim, C-S. (2018). Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement. In V. Ferrari, C. Sminchisescu, Y. Weiss, & M. Hebert (Eds.), Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings (pp. 39-55). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11208 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-01225-0_3

Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement. / Heo, Minhyeok; Lee, Jaehan; Kim, Kyung Rae; Kim, Han Ul; Kim, Chang-Su.

Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings. ed. / Vittorio Ferrari; Cristian Sminchisescu; Yair Weiss; Martial Hebert. Springer Verlag, 2018. p. 39-55 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11208 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Heo, M, Lee, J, Kim, KR, Kim, HU & Kim, C-S 2018, Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement. in V Ferrari, C Sminchisescu, Y Weiss & M Hebert (eds), Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11208 LNCS, Springer Verlag, pp. 39-55, 15th European Conference on Computer Vision, ECCV 2018, Munich, Germany, 18/9/8. https://doi.org/10.1007/978-3-030-01225-0_3
Heo M, Lee J, Kim KR, Kim HU, Kim C-S. Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement. In Ferrari V, Sminchisescu C, Weiss Y, Hebert M, editors, Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings. Springer Verlag. 2018. p. 39-55. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-01225-0_3
Heo, Minhyeok ; Lee, Jaehan ; Kim, Kyung Rae ; Kim, Han Ul ; Kim, Chang-Su. / Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement. Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings. editor / Vittorio Ferrari ; Cristian Sminchisescu ; Yair Weiss ; Martial Hebert. Springer Verlag, 2018. pp. 39-55 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{5a6261f26e7b4addadb26f98cf4e314d,
title = "Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement",
abstract = "We propose a monocular depth estimation algorithm based on whole strip masking (WSM) and reliability-based refinement. First, we develop a convolutional neural network (CNN) tailored for the depth estimation. Specifically, we design a novel filter, called WSM, to exploit the tendency that a scene has similar depths in horizonal or vertical directions. The proposed CNN combines WSM upsampling blocks with a ResNet encoder. Second, we measure the reliability of an estimated depth, by appending additional layers to the main CNN. Using the reliability information, we perform conditional random field (CRF) optimization to refine the estimated depth map. Experimental results demonstrate that the proposed algorithm provides the state-of-the-art depth estimation performance.",
keywords = "Depth map refinement, Monocular depth estimation, Reliability, Whole strip masking",
author = "Minhyeok Heo and Jaehan Lee and Kim, {Kyung Rae} and Kim, {Han Ul} and Chang-Su Kim",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-01225-0_3",
language = "English",
isbn = "9783030012243",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "39--55",
editor = "Vittorio Ferrari and Cristian Sminchisescu and Yair Weiss and Martial Hebert",
booktitle = "Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings",

}

TY - GEN

T1 - Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement

AU - Heo, Minhyeok

AU - Lee, Jaehan

AU - Kim, Kyung Rae

AU - Kim, Han Ul

AU - Kim, Chang-Su

PY - 2018/1/1

Y1 - 2018/1/1

N2 - We propose a monocular depth estimation algorithm based on whole strip masking (WSM) and reliability-based refinement. First, we develop a convolutional neural network (CNN) tailored for the depth estimation. Specifically, we design a novel filter, called WSM, to exploit the tendency that a scene has similar depths in horizonal or vertical directions. The proposed CNN combines WSM upsampling blocks with a ResNet encoder. Second, we measure the reliability of an estimated depth, by appending additional layers to the main CNN. Using the reliability information, we perform conditional random field (CRF) optimization to refine the estimated depth map. Experimental results demonstrate that the proposed algorithm provides the state-of-the-art depth estimation performance.

AB - We propose a monocular depth estimation algorithm based on whole strip masking (WSM) and reliability-based refinement. First, we develop a convolutional neural network (CNN) tailored for the depth estimation. Specifically, we design a novel filter, called WSM, to exploit the tendency that a scene has similar depths in horizonal or vertical directions. The proposed CNN combines WSM upsampling blocks with a ResNet encoder. Second, we measure the reliability of an estimated depth, by appending additional layers to the main CNN. Using the reliability information, we perform conditional random field (CRF) optimization to refine the estimated depth map. Experimental results demonstrate that the proposed algorithm provides the state-of-the-art depth estimation performance.

KW - Depth map refinement

KW - Monocular depth estimation

KW - Reliability

KW - Whole strip masking

UR - http://www.scopus.com/inward/record.url?scp=85055438092&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055438092&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-01225-0_3

DO - 10.1007/978-3-030-01225-0_3

M3 - Conference contribution

SN - 9783030012243

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 39

EP - 55

BT - Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings

A2 - Ferrari, Vittorio

A2 - Sminchisescu, Cristian

A2 - Weiss, Yair

A2 - Hebert, Martial

PB - Springer Verlag

ER -