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

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

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

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2018

https://arxiv.org/pdf/1804.02379.pdf

Contact: changhashin@yonsei.ac.kr

Environments

• Python3.5.2, Anaconda 4.2.0 (64-bit), Tensorflow 1.6.0 - 1.12.0
• pip install imageio

Train the EPINET

First, you need to download HCI Light field dataset from http://hci-lightfield.iwr.uni-heidelberg.de/. Unzip the LF dataset and move 'additional/, training/, test/, stratified/ ' into the 'hci_dataset/'.

And run python EPINET_train.py

• Checkpoint files will be saved in 'epinet_checkpoints/EPINET_train_ckp/iterXXX_XX.hdf5', it could be used for test EPINET model.
• Training process will be saved 'epinet_output/EPINET_train/train_XX.jpg'. (XX is iteration number).
• You might be change the setting 'learning rate','patch_size' and so on to get better result.

Test the EPINET

Run python EPINET_plusX_9conv22_save.py

• To test your own trained model from python EPINET_train.py, you need to modify the line 141-142 like below path_weight='epinet_checkpoints/EPINET_train_ckp/iter0097_trainmse2.706_bp12.06.hdf5'

Last modified date: 11/29/2018