def get_batch_with_label(self, N=None):
if N is None:
N = self.batch_size
batch = shuf(self.img_list, random_state=np.random.randint(1, 10000))
X = []
b = []
for i in range(N):
X.append(self.LoadImage(osp.join(self.source_folder, batch[i])))
b.append(
self.LoadImage(osp.join(self.source_folder, "blur", batch[i])))
X = np.array(X).reshape(N, -1)
b = np.array(b).reshape(N, -1)
return X, b
def get_batch(self, N=None):
if N is None:
N = self.batch_size
if N > len(self.img_list) or N == -1:
N = len(self.img_list)
batch = shuf(self.img_list, random_state=np.random.randint(1, 10000))
X = []
for i in range(N):
X.append(self.LoadImage(osp.join(self.source_folder, batch[i])))
X = np.array(X).reshape(N, -1)
z = np.zeros([N, self.K])
return X, z, z, self.lmbd
def get_batch_with_label(self, N=None, shuffle=True):
from keras.utils import np_utils
if N is None:
N = self.batch_size
im = self.im
if shuffle == True:
batch, label = shuf(im,
self.label,
random_state=np.random.randint(1, 10000))
else:
batch, label = im, self.label
X = batch[:N].reshape(N, -1)
label = label[:N]
label = np_utils.to_categorical(label, 10)
return X, label
def get_feature_batch(self,N):
from sklearn.utils import shuffle as shuf
batch,label = shuf(self.feat,self.label,random_state=np.random.randint(1,10000))
batch = batch[:N].reshape(N,-1)
label = label[:N]
return batch,label