/ZLFSR

Primary LanguagePython

ZLFSR

################################################################

This Project is a Pytorch implementation of "Learning Zero-shot dense light field reconstruction from heterogeneous imaging"

Requirements

** Python 3.7

** PyTorch 1.2.0

** CUDA 10.0

** Cudnn

** Matlab 2017b (for training/test data generation and evaluation)

Usage

** For simulation data

  1. Dataset Please download the dataset from http://lightfields.stanford.edu/mvlf/ (i.e., 3vlf_lfr.tar)

  2. Data generation Run ZLFSR_simulation/matlab/PrepareData.m to generate training data.

  3. Training Run ZLFSR_simulation/train.py to perform network training.

  4. Test Run ZLFSR_simulation/test.py to perform network testing.

** For real data

  1. Dataset Please download the dataset via Baidu Drive：https://pan.baidu.com/s/1Mekzz8M8Y7A1QSDwpxW-Lw (key: djrm)

  2. Data generation Run ZLFSR_real/matlab/PrepareRealData.m to generate training data.

  3. Training Run ZLFSR_real/train.py to perform network training.

  4. Test Run ZLFSR_real/test.py to perform network testing.

Citation

If you use this code for your research, please cite our paper.

Contact

If you have any questions about the code, please contact to cyy941027@126.com.