Inplementing the paper 'Context Encoders: Feature Learning by Inpainting' by TensorFlow
This code mainly inplement the paper Context Encoders: Feature Learning by Inpainting, and we use the dataset of cats' head for training and testing.
Cat dataset: http://academictorrents.com/details/c501571c29d16d7f41d159d699d0e7fb37092cbd Download the cat dataset, then unzip it and put them into the file 'cat_dataset'. Finally, execute the file 'prepocess_cat_dataset.py' to crop the cats' head.
├── cat_dataset
├── 00000001_000.jpg
├── 00000001_000.jpg.cat
├── 00000001_001.jpg
├── 00000001_001.jpg.cat
...
├── cats_bigger_than_64x64
├── 00000001_000.jpg
├── 00000001_001.jpg
├── 00000001_002.jpg
├── 00000001_003.jpg
...
├── cats_bigger_than_128x128
├── 00000001_000.jpg
├── 00000001_001.jpg
├── 00000001_002.jpg
├── 00000001_003.jpg
...
- tensorflow-gpu1.4.0
- numpy
- scipy
- pillow
- cv2
- glob
cat | 1 | 2 | 3 |
---|---|---|---|
a | |||
b | |||
c | |||
d | |||
e | |||
f | |||
g | |||
h |
RelativisticGAN-AlexiaJM