def main():
import sys
sys.path.remove('/opt/ros/kinetic/lib/python2.7/dist-packages'
) # in order to import cv2 under python3
import cv2
sys.path.append('/opt/ros/kinetic/lib/python2.7/dist-packages'
) # append back in order to import rospy
# parse arguments
args = parse_args()
if args is None:
exit()
# open session
tf_config = tf.ConfigProto()
tf_config.gpu_options.allow_growth = True
tf_config.allow_soft_placement = True
with tf.Session(config=tf_config) as sess:
gan = MUNIT(sess, args)
# build graph
gan.build_model()
# show network architecture
show_all_variables()
if args.phase == 'train':
# launch the graph in a session
gan.train()
print(" [*] Training finished!")
if args.phase == 'test':
gan.test()
print(" [*] Test finished!")
if args.phase == 'guide':
gan.style_guide_test()
print(" [*] Guide finished!")
if args.phase == 'opencv':
edge = np.zeros((256, 256, 3), dtype=np.uint8)
sketcher = Sketcher('image', [edge], lambda:
((255, 255, 255), 255))
while True:
key = cv2.waitKey()
if key == ord('q'):
exit()
if key == ord('r'):
edge[:] = np.zeros((256, 256, 3), dtype=np.uint8)
sketcher.show()
if key == 32: # hit spacebar
break
results = gan.test_opencv(edge)
print(" [*] Test finished!")
cv2.imshow('result', results)
cv2.waitKey(0)
import os, sys
MODE = 'remove' # 'remove', 'protect'
img_path = sys.argv[1]
def nothing(x):
pass
img = cv2.imread(img_path, cv2.IMREAD_COLOR)
img_masked = img.copy()
mask = np.zeros(img.shape[:2], np.uint8)
sketcher = Sketcher('image', [img_masked, mask], lambda:
((255, 255, 255), 255))
cv2.createTrackbar('width', 'image', img.shape[1], img.shape[1] * 2, nothing)
cv2.createTrackbar('height', 'image', img.shape[0], img.shape[0] * 2, nothing)
while True:
key = cv2.waitKey()
if key == ord('q'): # quit
break
if key == ord('r'): # reset
print('reset')
img_masked[:] = img
mask[:] = 0
sketcher.show()
if key == 32: # hit spacebar
from VideoAnalyzer import VideoAnalyzer
from Sketcher import Sketcher
import cv2
# input
model = cv2.imread('res/input/target.jpg')
video_name = 'res/input/video.mp4'
bullseye_point = (325,309)
inner_diameter_px = 50
inner_diameter_inch = 1.5
rings_amount = 6
display_in_cm = False
# get a sample frame from the video
cap = cv2.VideoCapture(video_name)
_, test_sample = cap.read()
# calculate the sizes of the frame and the input
model_h, model_w, _ = model.shape
frame_h, frame_w, _ = test_sample.shape
pixel_to_inch = inner_diameter_inch / inner_diameter_px
pixel_to_cm = pixel_to_inch * 2.54
measure_unit = pixel_to_cm if display_in_cm else pixel_to_inch
measure_unit_name = 'cm' if display_in_cm else '"'
# analyze
sketcher = Sketcher(measure_unit, measure_unit_name)
video_analyzer = VideoAnalyzer(video_name, model, bullseye_point, rings_amount, inner_diameter_px)
video_analyzer.analyze('res/output/output.mp4', sketcher)
# create the video capture.
video_capture = cv2.VideoCapture(0)
# initialize the correlation tracker.
tracker = dlib.correlation_tracker()
# this is the variable indicating whether to track the object or not.
tracked = False
cap = cv2.VideoCapture("video5.mp4")
ret, image = cap.read()
img_masked = image.copy()
mask = np.zeros(image.shape[:2], np.uint8)
point = []
sketcher=Sketcher('image', [img_masked, mask], lambda : ((255, 255, 255), 255), point)
cap.release()
# saving all points and find largest rectangle
while True:
key = cv2.waitKey()
if key == 32:
print('processing')
break
max_x = 0
max_y = 0
min_x = 9999
min_y = 9999
for pt in point:
if max_x < pt[0] : max_x = pt[0]
if max_y < pt[1] : max_y = pt[1]
print('load model...')
model = PConvUnet(vgg_weights=None, inference_only=True)
model.load('data/logs/MyDataset_phase1/weights.75-0.51.h5', train_bn=False)
# model.summary()
import os
src = 'D:\git\crawler\zigbang\\test\class\\'
dest = 'images\\'
def random_pick(path):
return os.listdir(path)[np.random.randint(0,len(os.listdir(path)))]
img = cv2.imread(os.path.join(src,random_pick(src)), cv2.IMREAD_COLOR)
img_masked = img.copy()
mask = np.zeros(img.shape[:2], np.uint8)
sketcher = Sketcher('image', [img_masked, mask], lambda : ((255, 255, 255), 255))
chunker = ImageChunker(512, 512, 30)
while True:
key = cv2.waitKey()
if key == ord('s'): #save
sketcher.save_files(dest)
if key == ord('c'):
new_img = cv2.imread(os.path.join(src, random_pick(src)), cv2.IMREAD_COLOR)
new_mask = np.zeros(new_img.shape[:2],np.uint8)
sketcher.dests[0] = new_img
sketcher.dests[1] = new_mask
sketcher.show()
if key == ord('q'): # quit
break
if key == ord('r'): # reset