def normalize_columns_together( headers,d ): #normalizes all the data so the min maps to 0 and the max maps to 1 data = d.get_data(headers) if data.size > 0: norm = data - data.min() if data.max() != 0: norm = norm/data.max() return norm
def highlight_max(data, color='#0e5c99'):
'''
highlight the maximum in a Series or DataFrame
'''
attr = 'background-color: {}'.format(color)
if data.ndim == 1: # Series from .apply(axis=0) or axis=1
is_max = data == data.max()
return [attr if v else '' for v in is_max]
else: # from .apply(axis=None)
is_max = data == data.max().max()
return pd.DataFrame(np.where(is_max, attr, ''),
index=data.index, columns=data.columns)
def visualize_data(data,
padsize=1,
padval=0,
cmap="gray",
image_size=(10, 10)):
data -= data.min()
data /= data.max()
# force the number of filters to be square
n = int(np.ceil(np.sqrt(data.shape[0])))
padding = ((0, n**2 - data.shape[0]), (0, padsize),
(0, padsize)) + ((0, 0), ) * (data.ndim - 3)
data = np.pad(data,
padding,
mode='constant',
constant_values=(padval, padval))
# tile the filters into an image
data = data.reshape((n, n) + data.shape[1:]).transpose(
(0, 2, 1, 3) + tuple(range(4, data.ndim + 1)))
data = data.reshape((n * data.shape[1], n * data.shape[3]) +
data.shape[4:])
#plt.figure(figsize=image_size)
plt.imshow(data, cmap=cmap)
plt.show()
plt.axis('off')
def generateDataImage(data,metadata,imgname):
fig = plt.figure(figsize=(8,4))
ax = fig.add_subplot(121)
ax.imshow(data,origin='lower',cmap=cm.jet)
ax2 = fig.add_subplot(122)
vmax = data.max()
if vmax < 1: vmax=1
ax2.imshow(data,origin='lower',cmap=cm.jet,norm=LogNorm(vmin=1, vmax=vmax))
#plt.colorbar(data,ax=ax2,norm=LogNorm(vmin=1))
fig.savefig(imgname)
def draw_img_rgb(data):
size = 96
n = data.shape[0]
plt.figure(figsize=(n*2, 2))
data /= data.max()
cnt = 1
for idx in np.arange(n):
plt.subplot(1, n, cnt)
tmp = data[idx,:,:,:].transpose(1,2,0)
plt.imshow(tmp)
plt.tick_params(labelbottom="off")
plt.tick_params(labelleft="off")
cnt+=1
plt.show()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--epoch', type=int, default=200, help='# of Epochs')
parser.add_argument('--learn',
type=float,
default=1e-1,
help='learning rate')
parser.add_argument('--photo', type=int, default=1, help='photo index')
FLAGS = parser.parse_args()
NEpochs = FLAGS.epoch
learningRate = FLAGS.learn
x = tf.Variable(tf.zeros([1, imageSize], dtype=tf.float32))
y = FLAGS.photo
w = tf.placeholder(tf.float32, shape=(imageSize, NClasses))
b = tf.placeholder(tf.float32, shape=(NClasses))
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
new_saver = tf.train.import_meta_graph(model_dir + 'model.meta')
new_saver.restore(sess, tf.train.latest_checkpoint(model_dir))
graph = tf.get_default_graph()
weights = sess.run(graph.get_tensor_by_name("weights:0"))
biases = sess.run(graph.get_tensor_by_name("biases:0"))
prediction = tf.nn.softmax(tf.matmul(x, w) + b)
loss = 1 - prediction[0][y]
optimizer = tf.train.GradientDescentOptimizer(learningRate).minimize(
loss)
for epoch in range(NEpochs):
_, l = sess.run([optimizer, loss],
feed_dict={
w: weights,
b: biases
})
print("Epoch: %01d loss: %.4f" % (epoch + 1, l))
data = np.asarray(sess.run(x), dtype = np.float32).\
reshape([-1, imageHeight, imageWidth])[0]
data_rescaled = (255.0 / data.max() * data).astype(np.uint8)
plt.imshow(np.asarray(sess.run(x)).reshape(
[-1, imageHeight, imageWidth])[0],
cmap='gray')
# plt.savefig('inversion/test_'+str(y)+'.png')
plt.savefig('/home/ubuntu/test_' + str(y) + '.png')
def visualize_data(data, padsize=1, padval=0, cmap="gray", image_size=(10,10)):
data -= data.min()
data /= data.max()
# force the number of filters to be square
n = int(np.ceil(np.sqrt(data.shape[0])))
padding = ((0, n ** 2 - data.shape[0]), (0, padsize), (0, padsize)) + ((0, 0),) * (data.ndim - 3)
data = np.pad(data, padding, mode='constant', constant_values=(padval, padval))
# tile the filters into an image
data = data.reshape((n, n) + data.shape[1:]).transpose((0, 2, 1, 3) + tuple(range(4, data.ndim + 1)))
data = data.reshape((n * data.shape[1], n * data.shape[3]) + data.shape[4:])
#plt.figure(figsize=image_size)
plt.imshow(data, cmap=cmap)
plt.show()
plt.axis('off')