CelebA 데이터셋 다운로드 하는법
Reference
- https://github.com/tkarras/progressive_growing_of_gans/tree/original-theano-version
- https://github.com/willylulu/celeba-hq-modified
CelebA 데이터셋
- 202,599 number of face images of various celebrities
- 10,177 unique identities, but names of identities are not given
- 40 binary attribute annotations per image
- 5 landmark locations
깃허브: https://github.com/jjxxmiin/CelebA_HQ_Downloader
만약 위에 깃허브를 clone하면 1번
2번
3번
생략
celeba-hq-modified
clone 하기
1
git clone https://github.com/willylulu/celeba-hq-modified.git
progressive_growing_of_gans
clone 하기
original-theano-version
이 필요한데 직접 clone이 되지 않아 download해서 압축을 풀어서 사용했음
page: https://github.com/tkarras/progressive_growing_of_gans/tree/original-theano-version
h5tool.py
바꾸기
1
cp celeb-hq-modified/h5tool.py progressive_growing_of_gans/h5tool.py
- 폴더 구성하기
1
2
3
4
5
6
7
8
cd progressive_growing_of_gans
mkdir celeba-hq
cd celeba-hq
mkdir celeba-64
mkdir celeba-128
mkdir celeba-256
mkdir celeba-512
mkdir celeba-1024
- 데이터 다운로드
- CelebA: https://drive.google.com/drive/folders/0B7EVK8r0v71peklHb0pGdDl6R28
- CelebA-HQ(delta): https://drive.google.com/drive/folders/0B4qLcYyJmiz0TXY1NG02bzZVRGs
CelebA
압축해제하기
1
7za e img_celeba.7z.001
- 데이터 전처리를 위한 폴더 구성
1
2
3
4
5
6
7
8
9
10
11
12
13
celebA/
A/
img_celeba/
Anno/
B/
deltas00000.zip
...
...
progressive_growing_of_gans/
h5tool.py
...
- 가상환경 만들기
1
2
3
conda create --name python_2 python=2
source activate python_2
pip install scipy numpy Pillow cryptography
- 시작 ^_^
1
2
cd progressive_growing_of_gans
python h5tool.py create_celeba_hq 123456.h5 ../celebA/A ../celebA/B
This post is licensed under CC BY 4.0 by the author.