def init(self):
self.camera = cv2.VideoCapture(
gstreamer_pipeline(1280, 720, 1280 // 2, 720 // 2, 20, 0),
cv2.CAP_GSTREAMER)
while True:
hasFrame, frame = self.camera.read()
if hasFrame:
Space.write(self.nameImage, frame, 0.15)
def __init__(self):
self.cap = cv2.VideoCapture(gstreamer_pipeline(flip_method=2),
cv2.CAP_GSTREAMER)
self.frame = None
self.stopped = False
self.success = False
print("Initializing Camera Thread")
speed = 70
def updateSpeed(*args):
global speed
speed = args[0]
robot.setSpeed(speed)
cv2.namedWindow("Circle")
cv2.createTrackbar("threshold", "Circle", threshold, 512, updateThreshold)
cv2.createTrackbar("speed", "Circle", speed, 255, updateSpeed)
camera = cv2.VideoCapture(
gstreamer_pipeline(1280, 720, 1280 // 2, 720 // 2, 20, 0),
cv2.CAP_GSTREAMER)
trials = 0
while True:
hasFrame, frame = camera.read()
if not hasFrame:
break
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
frame = cv2.blur(frame, (5, 5))
circles = cv2.HoughCircles(frame,
cv2.HOUGH_GRADIENT,
2,
def multi_style(path,
width=320,
device=device,
cycle_length=np.inf,
half_precision=False,
rotate=0,
camera=0,
cutoff=0):
if path.is_file():
model_iter = itertools.cycle([os.path.basename(path)])
path = os.path.dirname(path)
else:
model_iter = itertools.cycle(os.listdir(path))
model_file = next(model_iter)
print(f'Using {model_file} ')
model_path = os.path.join(path, model_file)
model = TransformerNet()
model.load_state_dict(read_state_dict(model_path))
model.to(device)
if half_precision:
model.half()
# attempts to load jetcam for Jetson Nano, if fails uses normal camera.
if rotate != 0:
width = int(width / .75)
height = int(width * .75)
if camera < 0:
#from jetcam.csi_camera import CSICamera
#vs = CSICamera(width=width, height=int(width/1.5), capture_width=1080, capture_height=720, capture_fps=15)
#vs.read()
print('Using CSI camera')
vs = cv2.VideoCapture(
gstreamer_pipeline(capture_width=width,
capture_height=height,
display_width=width,
display_height=height), cv2.CAP_GSTREAMER)
time.sleep(2.0)
img = vs.read()
assert img[1] is not None
else:
print('Using USB camera')
vs = VideoStream(src=camera, resolution=(width, height)).start()
time.sleep(2.0)
if rotate != 0:
width = int(width * .75)
timer = Timer()
cycle_begin = time.time()
while (True):
frame = vs.read()
if frame is None:
frame = np.random.randint(0,
255, (int(width / 1.5), width, 3),
dtype=np.uint8)
if type(frame) is type(()):
frame = frame[1]
frame = cv2.flip(frame, 1)
frame = resize(frame, width=width)
# Style the frame
img = style_frame(frame, model, device, half_precision).numpy()
img = np.clip(img, 0, 255)
img = img.astype(np.uint8)
img = img.transpose(1, 2, 0)
img = img[:, :, ::-1]
# rotate
if rotate != 0:
h, w, _ = img.shape
margin = int(w - h * h / w) // 2
img = img[:, margin:-margin, :]
if cutoff > 0:
margin = int(cutoff * img.shape[1]) // 2
img = img[:, margin:-margin, :]
elif cutoff < 0:
margin = int(-cutoff * img.shape[0]) // 2
img = img[margin:-margin, :, :]
# print(img.shape)
cv2.imshow("Output", img)
timer()
key = cv2.waitKey(1) & 0xFF
if key == ord("n") or (time.time() - cycle_begin) > cycle_length:
model_file = next(model_iter)
print(f'Using {model_file} ')
model_path = os.path.join(path, model_file)
model.load_state_dict(read_state_dict(model_path))
model.to(device)
cycle_begin = time.time()
elif key == ord("q"):
break
def photo_booth(path,
models,
width=1080,
device=torch.device('cpu'),
prep_time=10,
view_time=10,
rotate=0,
camera=0,
output_path=None,
cutoff=0):
if rotate != 0:
width = int(width / .75)
# attempts to load jetcam for Jetson Nano, if fails uses normal camera.
height = int(width * .75)
if camera < 0:
print('Using CSI camera')
vs = cv2.VideoCapture(
gstreamer_pipeline(capture_width=width,
capture_height=height,
display_width=width,
display_height=height), cv2.CAP_GSTREAMER)
time.sleep(2.0)
img = vs.read()
assert img[1] is not None
else:
print('Using USB camera')
vs = VideoStream(src=camera, resolution=(width, height)).start()
print('Warming up')
time.sleep(2.0)
if rotate != 0:
width = int(width * .75)
print('Program started')
model = TransformerNet()
cv2.namedWindow("Output", cv2.WND_PROP_FULLSCREEN)
cv2.setWindowProperty("Output", cv2.WND_PROP_FULLSCREEN,
cv2.WINDOW_FULLSCREEN)
while True:
preparation(vs, prep_time, rotate, cutoff)
img = vs.read()
if type(img) == type(()):
img = img[1]
img = cv2.flip(img, 1)
img = resize(img, width)
# rotate
if rotate != 0:
h, w, _ = img.shape
margin = int(w - h * h / w) // 2
img = img[:, margin:-margin, :]
if cutoff > 0:
margin = int(cutoff * img.shape[1]) // 2
img = img[:, margin:-margin, :]
elif cutoff < 0:
margin = int(-cutoff * img.shape[0]) // 2
img = img[margin:-margin, :, :]
# Choosing and loading the model
model_name = np.random.choice(models)
print('Using {}'.format(model_name))
model_path = os.path.join(path, model_name)
state_dict = read_state_dict(model_path)
model.load_state_dict(state_dict)
model.to(device)
if output_path is not None:
if not os.path.exists(output_path):
os.mkdir(output_path)
filename = time.strftime('%Y_%m_%d_%H_%M_%S_') + os.path.splitext(
model_name)[0] + '.jpg'
filepath = os.path.join(output_path, filename)
cv2.imwrite(filepath, img)
# Inference
cv2.imshow('Output', img)
key = cv2.waitKey(1) & 0xFF
busy = BusyShow(img, style_frame, **{
'img': img,
'style_model': model,
'device': device
})
output = busy.execute()
output = output.numpy()
# output = style_frame(img,model,device).numpy()
# Postprocessing
output = post_process(output)
cv2.imshow('Output', output)
key = cv2.waitKey(1) & 0xFF
if key == ord("q"):
break
view_result(img, view_time)