def collect_one_point(camera, steering_angle, Xs, ys):
frame = camera.capture()
frame = resize(frame)
set_steering(steering_angle)
time.sleep(0.1)
set_throttle(CAR_THROTTLE_FORWARD_SAFE_SPEED)
time.sleep(0.2)
set_throttle(0.0)
Xs.append(frame)
ys.append(steering_angle)
def collect_one_point(camera, steering_angle, Xs, ys):
frame = camera.capture()
frame = resize(frame)
set_steering(steering_angle)
time.sleep(0.1)
set_throttle(safe_forward_throttle())
time.sleep(0.2)
set_throttle(0.0)
Xs.append(frame)
ys.append(steering_angle)
def fill_gaps(self):
contours = pp.get_all_rects(self.image)
avg_size = 0
for x in contours:
avg_size += cv2.contourArea(x[4])
avg_size /= len(contours)
for x in contours:
contour_size = cv2.contourArea(x[4])
if contour_size > (avg_size * 10):
self.image = cv2.drawContours(self.image, [x[4]], 0, 128, -1)
cv2.imshow("contours", pp.resize(self.image, 0.5))
cv2.waitKey(0)
def build_graph(self,
images_path,
image_name,
graph_polygon_path=None,
skeletons_path=None,
graph_grid_path=None,
graph_grid_adjacent_path=None):
image = imread(images_path + image_name)
image_resized = preprocessing.resize(
image, (self.image_height,
image.shape[1] * self.image_height // image.shape[0]))
image_binarized = preprocessing.binarize(image_resized, self.threshold)
image_inverted = preprocessing.invert(image_binarized)
image_skeletonized = preprocessing.skeletonize(image_inverted)
if skeletons_path is not None:
imsave(skeletons_path + image_name,
np.array((1 - image_skeletonized) * 255, dtype=np.uint8))
grp = GraphRepresentatorPolygon(image_inverted, self.polygon_tolerance,
image_name)
graph_image_polygon = grp.represent_graph_full()
if graph_grid_path is not None:
grg = GraphRepresentatorGrid(image_skeletonized, self.cell_length)
graph_image_grid = grg.represent_graph()
imsave(graph_grid_path + image_name,
np.array((1 - graph_image_grid) * 255, dtype=np.uint8))
if graph_grid_adjacent_path is not None:
grg = GraphRepresentatorGrid(image_skeletonized, self.cell_length)
graph_image_grid = grg.represent_graph_no_grid_connection()
imsave(graph_grid_adjacent_path + image_name,
np.array((1 - graph_image_grid) * 255, dtype=np.uint8))
if graph_polygon_path is not None:
imsave(graph_polygon_path + image_name,
np.array((1 - graph_image_polygon) * 255, dtype=np.uint8))
return grp
import cv2 as cv
import os
# pre-process test images
import preprocessing
test_dir = './images/all'
#test_dir = './images/new_test'
#test_dir = './images/bad_predict'
#test_dir = './images/test'
filenames = glob(os.path.join(test_dir, '*.png'))
for i, file in enumerate(filenames):
#print('processing:', file)
img = cv.imread(file)
resized = preprocessing.resize(img)
img_name = str(i) + '.png'
#filepath = os.path.join('./images/test_buf', img_name)
filepath = os.path.join('./images/all_resize', img_name)
cv.imwrite(filepath, resized)
"""
def load_test_images(file_list):
test_set = list()
test_set_rgb = list()
for file in file_list:
#print(file)
img = cv.imread(file)
img_rgb = cv.cvtColor(img, cv.COLOR_BGR2RGB)
test_set.append(img)
test_set_rgb.append(img_rgb)
import tensorflow as tf
import os
import preprocessing
from model import CNN_Model
preprocessing.make_folder() # Create a folder to save the resized image
preprocessing.resize() # Resize image to 64 by 64 and change it to grayscale
preprocessing.make_csv() # Create labeled image csv
# hyper parameter
IMAGE_WIDTH = 64
IMAGE_HEIGHT = 64
BATCH_SIZE = 125
NUM_CLASSES = 2
CHECK_POINT_DIR = TB_SUMMARY_DIR = "./tensor_board/"
# read csv
csv_file = tf.train.string_input_producer(["./label.csv"], shuffle=True)
csv_reader = tf.TextLineReader()
_, line = csv_reader.read(csv_file)
image_file, label_decoded = tf.decode_csv(line, record_defaults=[[""], [""]])
image_decoded = tf.image.decode_jpeg(tf.read_file(image_file), channels=1)
image_cast = tf.cast(image_decoded, tf.float32)
image = tf.reshape(image_cast,
[IMAGE_WIDTH, IMAGE_HEIGHT, 1]) # 64 by 64 , grayscale
test_batch = int(12500 / BATCH_SIZE)
test_image_list = [
'./resize_test/' + file_name for file_name in os.listdir('./resize_test/')
]