2017
DOI: 10.1007/978-3-319-70090-8_65
|View full text |Cite
|
Sign up to set email alerts
|

EPI-Patch Based Convolutional Neural Network for Depth Estimation on 4D Light Field

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
37
0

Year Published

2019
2019
2024
2024

Publication Types

Select...
3
3
1

Relationship

0
7

Authors

Journals

citations
Cited by 47 publications
(37 citation statements)
references
References 21 publications
0
37
0
Order By: Relevance
“…Recently, deep learning methods have gained much attention in estimating depth from light fields. Heber et al [14], Luo et al [15], Heber et al [16] and Feng et al [17] feed the input of Epipolar Plane Image (EPI) to the ConvNet where the network learns the proportional relation between the slope of the epipolar line and depth. However, this relation is hard to learn in wide-baseline light fields due to the absence of the epipolar line on the EPI.…”
Section: A Deep Learning-based Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…Recently, deep learning methods have gained much attention in estimating depth from light fields. Heber et al [14], Luo et al [15], Heber et al [16] and Feng et al [17] feed the input of Epipolar Plane Image (EPI) to the ConvNet where the network learns the proportional relation between the slope of the epipolar line and depth. However, this relation is hard to learn in wide-baseline light fields due to the absence of the epipolar line on the EPI.…”
Section: A Deep Learning-based Methodsmentioning
confidence: 99%
“…A light field image from the camera is usually separated into the so-called subaperture images, and the baseline between sub-aperture images is very narrow. To date, traditional [2][3][4][5][6][7][8][9][10][11][12][13] and ConvNet-based [14][15][16][17][18][19][20] methods have been well studied for high performance in narrow-baseline light fields, and achieved a low percentage of errors, e.g., EPINET [19]. For wide-baseline light fields, they are usually captured by a camera array or gantry (i.e., Manuscript received 2020.…”
Section: Introductionmentioning
confidence: 99%
“…Based on the epipolar plane image (EPI) or epipolar geometry property, [10] proposed to formulate the depth estimation as a classification problem, in which a standard CNN-architecture is employed on horizontal and vertical EPI patches. Since a shallow CNN is inadequate to guarantee the accuracy, a global optimization with traditional approach is utilized.…”
Section: Related Workmentioning
confidence: 99%
“…Light field depth estimation now turns into an active research topic with its potential to obtain more accurate depth maps [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16], which is important to many depth-based vision applications, such as 3D reconstruction, semantic segmentation. In these vision fields, deep learning approaches typically surpass the traditional methods, both in accuracy and speed.…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation