# Preprocessing for without shadow
gray1 = cv2.cvtColor(img, cv2.COLOR_BGRA2GRAY)
blur1 = cv2.medianBlur(gray1, 11)
ret, th3shadow = cv2.threshold(blur1, 100, 255, cv2.THRESH_BINARY)
th3 = cv2.medianBlur(th3shadow, 5)
# Canny edge detector
cannyed_image_shadow = cv2.Canny(th3, 10, 250)
# Preprocessing for canny with shadow
gray = cv2.cvtColor(img, cv2.COLOR_BGRA2GRAY)
blur = cv2.medianBlur(gray, 5)
cannyed_image = cv2.Canny(blur, 75, 250)
# Road only mask
if not os.path.isfile("output.npy"):
city_segment.segmentation("example.jpg", "Outputs/output.jpg")
mask = city_segment.largest_connected_component()
# Apply mask
canny_road1 = cv2.bitwise_and(cannyed_image_shadow, mask)
canny_road2 = cv2.bitwise_and(cannyed_image, mask)
# Subtraction
woshadow = canny_road2 - canny_road1
# Show
cv2.imshow("", woshadow)
cv2.waitKey(-1)
# Our libraries
import city_segment
import parking_detection
import vehicle_segment
if __name__ == "__main__":
filename = "example.jpg"
# High level segmentation using pre-trained network
if not os.path.isfile("output.npy"):
city_segment.segmentation(filename, "output.png")
# Create mask for only largest road segment
mask = city_segment.largest_connected_component(filename="output.npy")
#mask = city_segment.dilation(mask, 200)
# Apply traditional CV to masked image
img = cv2.imread(filename)
lines = parking_detection.get_lines(img, mask, [1, 0.001, 10, 60, 40])
#line_image = parking_detection.draw_lines(img, lines)
# Detect vehicles in image
car_lines = vehicle_segment.segmentation(filename)
line_image = parking_detection.draw_lines(img, car_lines)
# Post-processing on lines
h_lines = parking_detection.lines_processing(lines)
parking_detection.get_scatter(h_lines)
x, y, n = parking_detection.hc()