vae_size[1]),
axis=0)
pred = sess.run([model.z_mean, model.z_sigma],
feed_dict={
model.inputs['X']: img2encode,
model.trainable: False
})
encoded = np.zeros((1, model.latent_num * 2))
encoded[0, :model.latent_num] = pred[0][0, :]
encoded[0, model.latent_num:] = pred[1][0, :]
# compute llh
llh = np.zeros(5)
for i in range(5):
llh[i] = gmm_models[i].score_samples(encoded)[0]
llh_mask[i].append(llh[i] * np.ones(
(patch_size[0], patch_size[1], 1)))
# get mask
plt.figure(figsize=(15, 5))
for i in range(5):
conf_map = uabUtilreader.un_patchify(np.stack(llh_mask[i], axis=0),
tile_dim, patch_size, 0)
plt.subplot(151 + i)
plt.imshow(conf_map[:, :, 0])
plt.axis('off')
plt.tight_layout()
#plt.savefig(os.path.join(img_dir, '{}_{}_{}_gmm_all.png'.format(city_name, city_id, model_name)))
plt.show()
def run(self,
train_reader=None,
valid_reader=None,
test_reader=None,
pretrained_model_dir=None,
layers2load=None,
isTrain=False,
img_mean=np.array((0, 0, 0), dtype=np.float32),
verb_step=100,
save_epoch=5,
gpu=None,
tile_size=(5000, 5000),
patch_size=(572, 572),
truth_val=1,
continue_dir=None,
load_epoch_num=None,
fineTune=False,
valid_iou=False,
best_model=True):
if gpu is not None:
os.environ['CUDA_DEVICE_ORDER'] = 'PCI_BUS_ID'
os.environ['CUDA_VISIBLE_DEVICES'] = str(gpu)
if isTrain:
coord = tf.train.Coordinator()
with tf.Session(config=self.config) as sess:
# init model
init = [
tf.global_variables_initializer(),
tf.local_variables_initializer()
]
sess.run(init)
saver = tf.train.Saver(var_list=tf.global_variables(),
max_to_keep=1)
# load model
if pretrained_model_dir is not None:
if layers2load is not None:
self.load_weights(pretrained_model_dir, layers2load)
else:
if not fineTune:
restore_var = [
v for v in tf.global_variables() if
'resnet_v1' in v.name and not 'Adam' in v.name
]
loader = tf.train.Saver(var_list=restore_var)
self.load(pretrained_model_dir,
sess,
loader,
epoch=load_epoch_num)
else:
self.load(pretrained_model_dir,
sess,
saver,
epoch=load_epoch_num)
threads = tf.train.start_queue_runners(coord=coord, sess=sess)
try:
train_summary_writer = tf.summary.FileWriter(
self.ckdir, sess.graph)
self.train('X',
'Y',
self.n_train,
sess,
train_summary_writer,
n_valid=self.n_valid,
train_reader=train_reader,
valid_reader=valid_reader,
image_summary=util_functions.image_summary,
img_mean=img_mean,
verb_step=verb_step,
save_epoch=save_epoch,
continue_dir=continue_dir,
valid_iou=valid_iou)
finally:
coord.request_stop()
coord.join(threads)
saver.save(sess,
'{}/model.ckpt'.format(self.ckdir),
global_step=self.global_step)
else:
kernel = _window_2D(patch_size[0] + 182, patch_size[0])
with tf.Session() as sess:
init = tf.global_variables_initializer()
sess.run(init)
self.load(pretrained_model_dir,
sess,
epoch=load_epoch_num,
best_model=best_model)
self.model_name = pretrained_model_dir.split('/')[-1]
result = self.test('X', sess, test_reader)
n_patch = result.shape[0]
for patch_cnt in range(n_patch):
result[patch_cnt, :, :, :] *= kernel
image_pred = uabUtilreader.un_patchify(result, tile_size,
patch_size)
return image_pred
