dcgan-matconvnet-master
所属分类:matlab编程
开发工具:matlab
文件大小:4KB
下载次数:8
上传日期:2020-08-12 18:33:25
上 传 者:
6774839
说明: #在MatConvNet上实施DCGAN
MatConvNet上的深度卷积生成对抗网络(DCGAN注意:任何MCN版本均兼容
(# Implement DCGAN on MatConvNet
Deep Convolution Generative Adversarial Network on MatConvNet (DCGAN Note: Any MCN version is compatible)
文件列表:
get_test_DCGAN.m (1778, 2017-06-29)
get_train_DCGAN.m (6087, 2017-06-29)
main_start_dcgan.m (2022, 2017-06-29)
pax_global_header (52, 2017-03-24)
# DCGAN implementation on MatConvNet
Deep Convolutional Generative Adversarial Network (DCGAN) [1] implementation on MatConvNet
*Note: Any MCN version is compatible
## Prerequisites
1. Download CelebA database at: https://drive.google.com/file/d/0B7EVK8r0v71pZjFTYXZWM3FlRnM/view?usp=sharing
Unzip and store it to '/data' folder such that 'data/img_align_celeba/00001.bmp..'
2. Download Pretrained DCGAN at: https://www.dropbox.com/s/dmyqd6yfga9wsm6/net_dcgan.mat?dl=0
Place it in '/net' folder such that 'net/net_dcgan.mat'
## How to play
Run main_start_dcgan.m
If you use GPU, set opts.idx_gpus = 1' in line 7 and opts.install.cuda_path = 'your cuda folder' in line 11)
'main_start_dcgan.m' will perform:
1. Installation of MatConvNet (ver 1.0-beta24)
2. Generation of an image DB file fitting to MatConvNet
3. Testing a pre-trained DCGAN
4. Training a new DCGAN
*Note: All hyper parameters for training the network (e.g., learning rate) are identically set to [1]
## Results
![alt text](https://github.com/sunghbae/dcgan-matconvnet/blob/master/demo.png)
[1] Radford, Alec, Luke Metz, and Soumith Chintala. "Unsupervised representation learning with deep convolutional generative adversarial networks." arXiv preprint arXiv:1511.0***34 (2015).(https://arxiv.org/abs/1511.0***34)
近期下载者:
相关文件:
收藏者: