def _load_batch_data(self, indices):
filenames = [
os.path.join(self.celeba_path, f'{ind+1:06}.jpg')
for ind in indices
]
images = [
utils.get_image(name,
input_height=178,
input_width=218,
resize_height=self.res,
resize_width=self.res,
is_crop=True) / 255. for name in filenames
]
images = np.asarray(images)
return images
def _load_data(self):
TRAIN_STOP = 100000
data = json.load(open('/hdd_c/data/MITFaces/img_store'))[:TRAIN_STOP]
images = [
get_image('/hdd_c/data/MITFaces/notebooks/' + name,
input_height=178,
input_width=218,
resize_height=64,
resize_width=64,
is_crop=False) / 255. for name in data
]
print('finish reading face images')
images = np.asarray(images)
print('finish converting face images to numpy array')
factors = np.load('../../MITFaces/z_store_full')[:TRAIN_STOP]
print('finish reading factors')
return images, factors
def _load_data(self):
TRAIN_STOP = self.num_samples
self._TRAIN_STOP = TRAIN_STOP
print(TRAIN_STOP)
celebA = [
os.path.join(root, filename)
for root, dirnames, filenames in os.walk(self.celeba_path)
for filename in filenames if filename.endswith('.jpg')
]
celebA = celebA[:TRAIN_STOP]
images = [
utils.get_image(name,
input_height=178,
input_width=218,
resize_height=64,
resize_width=64,
is_crop=True) / 255. for name in celebA
]
print('finish reading face images')
images = np.float32(np.asarray(images))
print(images.shape)
return images