sess.run(init)
# restore previous model if there is one
ckpt = tf.train.get_checkpoint_state(CHECKPOINT_DIR)
if ckpt and ckpt.model_checkpoint_path:
print "Restoring previous model..."
try:
saver.restore(sess, ckpt.model_checkpoint_path)
print "Model restored"
except:
print "Could not restore model"
raise
exit()
print 'Loading data...'
images, annots, test_images, test_annots, _ = data_ops.load_galaxy(
DATA_DIR)
test_len = len(test_annots)
print 'generating data...'
idx = np.random.choice(np.arange(test_len), BATCH_SIZE, replace=False)
batch_y = test_annots[idx]
# the four z vectors to interpolate between
f_z = np.random.normal(-1.0, 1.0, size=[4, 100]).astype(np.float32)
# contains rows of images
gen_imgs = []
y1 = batch_y[0]
y2 = batch_y[1]
y3 = batch_y[2]
y4 = batch_y[3]
# restore previous model if there is one
ckpt = tf.train.get_checkpoint_state(CHECKPOINT_DIR)
if ckpt and ckpt.model_checkpoint_path:
print "Restoring previous model..."
try:
saver.restore(sess, ckpt.model_checkpoint_path)
print "Model restored"
except:
print "Could not restore model"
pass
print 'Loading data...'
# images and annots: _, __
train_images, train_annots, test_images, test_annots, paths = data_ops.load_galaxy(
DATA_DIR)
test_len = len(test_annots)
print 'test num:', test_len
info = {}
# want to write out a file with the image path and z vector
for p, img, label in tqdm(zip(paths, test_images, test_annots)):
img = data_ops.normalize(img)
batch_images = np.expand_dims(img, 0)
encoding = sess.run([encoded], feed_dict={images: batch_images})[0][0]
info[p] = [encoding, label]
# write out dictionary to pickle file
p = open(OUTPUT_DIR + 'data.pkl', 'wb')