×

Algorithm Details

Check out the new Google group to connect with other light field researchers, discuss ideas, and initiate collaborations. Some questions about the benchmark are answered over there as well! There is also a new collection of light field resources. Please feel encouraged to add further links.
Acronym

Epinet-fcn9x9

Method title

Epinet - fully convolutional network

Method description

Fully convolutional network for light field image.
- Input : star shape in 9x9 views(33views), RGB images
- 22 convolutional layers.




------------------------------------------------------------------------------------------
Epinet-fcn: star shape in 7x7 views(25views)

Camera setup

star shape in 9x9 views

Parameter description

total 22 convolutional layers.
convolutional filter num (70,70,..70,280,280... 280)

Programming language

python

Runtime environment

Window10 64bit, i7-7700 @3.6GHz, 32GB RAM, 1080ti

Additional data

Source code

Publication

Title

EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth from Light Field Images

Authors

Changha Shin, Hae-Gon Jeon, Youngjin Yoon, In So Kweon, Seon Joo Kim

Conference

CVPR'18

BibTex

@inproceedings{shin18epinet,
title={EPINET: A Fully-Convolutional Neural Network Using Epipolar Geometry for Depth from Light Field Images },
author={Changha Shin, Hae-Gon Jeon, Youngjin Yoon, In So Kweon, Seon Joo Kim},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
year={2018}
}