pydemia/hd-celeba-cropper

Obtain high resolution face images from CelebA

HD CelebA Cropper

CelebA dataset provides an aligned set img_align_celeba.zip. However, the size of each aligned image is 218x178, so the faces cropped from such images would be even smaller!

Here we provide a code to obtain higher resolution face images, by cropping the faces from the original unaligned images via 5 landmarks.

We also use a deep image quality assessment method to evaluate and rank the cropped image quality in scores.txt, lower score the better.

Cropped Faces (512x512)

Notice: There are still some low resolution cropped faces since the corresponding original images are low resolution.

Usage

• Prerequisites

  • OpenCV 3 (much faster) or scikit-image
  • Python 2 or 3

• Dataset

  • download the dataset

    • img_celeba.7z
      • https://pan.baidu.com/s/1eSNpdRG#list/path=%2FCelebA%2FImg or
      • https://drive.google.com/drive/folders/0B7EVK8r0v71pTUZsaXdaSnZBZzg
    • list_landmarks_celeba.txt
      • https://pan.baidu.com/s/1eSNpdRG#list/path=%2FCelebA%2FAnno&parentPath=%2F or
      • https://drive.google.com/drive/folders/0B7EVK8r0v71pOC0wOVZlQnFfaGs

  • unzip the data

    7z x img_celeba.7z.001

  • organize the files as follow

    path_to_dataset
    ├── data/*.jpg
    └── list_landmarks_celeba.txt
    

• Examples

  • 512x512 + lanczos4 (with OpenCV) + jpg

    python hd_celeba.py --data_dir path_to_dataset --crop_size 512 --order 4 --save_format jpg --n_worker 32

    

  • 512x512 + lanczos4 (with OpenCV) + png + larger face in the image (by setting face_factor, default is 0.7)

    python hd_celeba.py --data_dir path_to_dataset --crop_size 512 --order 4 --save_format png --face_factor 0.8 --n_worker 32

    

  • 384x384 + bicubic + jpg + smaller face in the image (by setting face_factor, default is 0.7)

    python hd_celeba.py --data_dir path_to_dataset --crop_size 384 --order 3 --save_format jpg --face_factor 0.65 --n_worker 32

    

• Notice

  • order for OpenCV
    • 0: INTER_NEAREST
    • 1: INTER_LINEAR
    • 2: INTER_AREA
    • 3: INTER_CUBIC
    • 4: INTER_LANCZOS4
    • 5: INTER_LANCZOS4
  • order for scikit-image
    • 0: Nearest-neighbor
    • 1: Bi-linear
    • 2: Bi-quadratic
    • 3: Bi-cubic
    • 4: Bi-quartic
    • 5: Bi-quintic