def main(input1):
input1=doPasing()
testDataset = cache(cache_path='my_dataset_cache_test.pkl',
fn=testDataset,
input_dir=input1.input_dir)
imagesOfTests = testDataset.images
graph = tf.Graph()
with graph.as_default():
with tf.Session() as sess:
#import the model dir
try:
file_=Path(input1.meta_file)
abs_path=file_.resolve()
except FileNotFoundError:
sys.exit('Meta File Not found')
else:
imported_meta = tf.train.import_meta_graph(input1.meta_file)
if os.path.isdir(input1.chk_point_dir):
imported_meta.restore(sess, tf.train.latest_checkpoint(input1.chk_point_dir))
for i in range(0,h_no):
for j in range(0,):
a = predictionOfM[i,j]
theOutput[128*i:128*(i+1),128*j:128*(j+1),:] = 1-a
newImages = image[0:h_no*128,0:*128,:]
prediction = np.multiply(theOutput,newImages)
else:
def main(args):
#File names are saved into a cache file
args = parse_arguments()
dataset_test = cache(cache_path='my_dataset_cache_test.pkl',
fn=Dataset_test,
in_dir=args.in_dir)
test_images = dataset_test.images
graph = tf.Graph()
with graph.as_default():
with tf.Session() as sess:
#import the model dir
try:
file_ = Path(args.meta_file)
abs_path = file_.resolve()
except FileNotFoundError:
sys.exit('Meta File Not found')
else:
imported_meta = tf.train.import_meta_graph(args.meta_file)
if os.path.isdir(args.chk_point_dir):
imported_meta.restore(
sess, tf.train.latest_checkpoint(args.chk_point_dir))
else:
sys.exit("Check Point Directory does not exist")
x = graph.get_operation_by_name("x").outputs[0]
predictions = graph.get_operation_by_name("predictions").outputs[0]
#Take one image at a time, pass it through the network and save it
for counter, image in enumerate(test_images):
broken_image, h, w, h_no, w_no = break_image(image, 128)
output_image = np.zeros((h_no * 128, w_no * 128, 3),
dtype=np.uint8)
feed_dict = {x: broken_image}
batch_predictions = sess.run(predictions, feed_dict=feed_dict)
matrix_pred = batch_predictions.reshape((h_no, w_no))
#Concentrate after this for post processing
for i in range(0, h_no):
for j in range(0, w_no):
a = matrix_pred[i, j]
output_image[128 * i:128 * (i + 1),
128 * j:128 * (j + 1), :] = 1 - a
cropped_image = image[0:h_no * 128, 0:w_no * 128, :]
pred_image = np.multiply(output_image, cropped_image)
print("Saved {} Image(s)".format(counter + 1))
cv2.imwrite(
os.path.join(args.save_dir,
'outfile_{}.jpg'.format(counter + 1)),
pred_image)
def main(args):
#File names are saved into a cache file
args=parse_arguments()
dataset_test = cache(cache_path='my_dataset_cache_test.pkl',
fn=Dataset_test,
in_dir=args.in_dir)
test_images = dataset_test.images
graph = tf.Graph()
with graph.as_default():
with tf.Session() as sess:
#import the model dir
try:
file_=Path(args.meta_file)
abs_path=file_.resolve()
except FileNotFoundError:
sys.exit('Meta File Not found')
else:
imported_meta = tf.train.import_meta_graph(args.meta_file)
if os.path.isdir(args.chk_point_dir):
imported_meta.restore(sess, tf.train.latest_checkpoint(args.chk_point_dir))
else:
sys.exit("Check Point Directory does not exist")
x = graph.get_operation_by_name("x").outputs[0]
predictions = graph.get_operation_by_name("predictions").outputs[0]
#Take one image at a time, pass it through the network and save it
for counter,image in enumerate(test_images):
broken_image,h,w,h_no,w_no = break_image(image,128)
output_image = np.zeros((h_no*128,w_no*128,3),dtype = np.uint8)
feed_dict = {x: broken_image}
batch_predictions = sess.run(predictions, feed_dict = feed_dict)
matrix_pred = batch_predictions.reshape((h_no,w_no))
#Concentrate after this for post processing
for i in range(0,h_no):
for j in range(0,w_no):
a = matrix_pred[i,j]
output_image[128*i:128*(i+1),128*j:128*(j+1),:] = 1-a
cropped_image = image[0:h_no*128,0:w_no*128,:]
pred_image = np.multiply(output_image,cropped_image)
print("Saved {} Image(s)".format(counter+1))
cv2.imwrite(os.path.join(args.save_dir,'outfile_{}.jpg'.format(counter+1)), pred_image)
def main(args):
#File names are saved into a cache file
args = parse_arguments()
dataset_test = cache(cache_path='my_dataset_cache_test.pkl',
fn=Dataset_test,
in_dir=args.in_dir)
test_images = dataset_test.images
nums = len(list(enumerate(test_images)))
graph = tf.Graph()
with graph.as_default():
with tf.Session() as sess:
#import the model dir
try:
file_ = Path(args.meta_file)
abs_path = file_.resolve()
except FileNotFoundError:
sys.exit('Meta File Not found')
else:
imported_meta = tf.train.import_meta_graph(args.meta_file)
if os.path.isdir(args.chk_point_dir):
imported_meta.restore(
sess, tf.train.latest_checkpoint(args.chk_point_dir))
else:
sys.exit("Check Point Directory does not exist")
x = graph.get_operation_by_name("x").outputs[0]
predictions = graph.get_operation_by_name("predictions").outputs[0]
leng = []
leng.append(nums)
#Take one image at a time, pass it through the network and save it
for counter, image in enumerate(test_images):
cnt = 0
broken_image, h, w, h_no, w_no = break_image(image, 128)
output_image = np.zeros((h_no * 128, w_no * 128, 3),
dtype=np.uint8)
feed_dict = {x: broken_image}
batch_predictions = sess.run(predictions, feed_dict=feed_dict)
matrix_pred = batch_predictions.reshape((h_no, w_no))
#Concentrate after this for post processing
for i in range(0, h_no):
for j in range(0, w_no):
a = matrix_pred[i, j]
output_image[128 * i:128 * (i + 1),
128 * j:128 * (j + 1), :] = 1 - a
if 1 - a == 1:
cnt += 1
cropped_image = image[0:h_no * 128, 0:w_no * 128, :]
pred_image = np.multiply(output_image, cropped_image)
print("Saved {} Image(s)".format(counter + 1))
leng.append(cnt * 128 * 1.414)
cv2.imwrite(
os.path.join(args.save_dir,
'outfile_{}.jpg'.format(counter + 1)),
pred_image)
index = 0
for img in os.listdir(
"./Save"): # read all the images in dir "Save"
# if not hidden folder
if not img.startswith('.'):
index += 1
# read the image img1
img1 = cv2.imread("./Save" + "/" + img,
cv2.IMREAD_GRAYSCALE)
# get the shape
size = img1.shape
lengthcnt = 0
widthcnt = 0
length = [0] * size[1]
width = [0] * size[0]
for i in range(size[0]):
for j in range(size[1]):
# if this pix is in the crack-zone
if img1[i][j] > 0 and length[j] == 0:
length[j] = 1
lengthcnt += 1
if img1[i][j] > 0 and width[i] == 0:
width[i] = 1
widthcnt += 1
# get the length of the crack
len_vec = leng
totLen = len_vec[len_vec[0] - index + 1]
print(
img +
"中裂缝的行投影为:%d像素，占图片长度的%.2f%%，裂缝的列投影为:%d像素，占图片高度的%.2f%%。裂缝的总长约为:%.3f像素"
% (lengthcnt, lengthcnt * 100.0 / size[1], widthcnt,
widthcnt * 100.0 / size[0], totLen))