def run(self):
VN_logger.text(VN_logger.GENERAL, "Running {0}".format(self.name()))
# start a video capture
if self.simulator:
VN_logger.text(VN_logger.GENERAL, "Using simulator")
sim.load_target(self.target_file, self.target_size)
sim.set_target_location(veh_control.get_home(True))
# sim.set_target_location(Location(0,0,0))
else:
VN_video.start_capture(self.camera_index)
# create an image processor
# detector = CircleDetector()
detector = TopCode()
# create a queue for images
imageQueue = []
# initilize autopilot variables
location = Location(0, 0, 0)
attitude = Attitude(0, 0, 0)
while veh_control.is_connected():
"""
#kill camera for testing
if(cv2.waitKey(2) == 1113938):
self.kill_camera = not self.kill_camera
"""
# we are in the landing zone or in a landing mode and we are still running the landing program
# just because the program is running does not mean it controls the vehicle
# i.e. in the landing area but not in a landing mode
# FIXME add inside_landing_area() back to conditional
if (self.always_run) or (veh_control.get_mode() == "LAND") or (veh_control.get_mode() == "RTL"):
# update how often we dispatch a command
VN_dispatcher.calculate_dispatch_schedule()
# update simulator
if self.simulator:
# get info from autopilot
location = veh_control.get_location()
attitude = veh_control.get_attitude()
sim.refresh_simulator(location, attitude)
# grab an image
frame = None
capStart = current_milli_time()
if self.simulator:
frame = sim.get_frame()
else:
frame = VN_video.get_image()
capStop = current_milli_time()
"""
if(self.kill_camera):
frame[:] = (0,255,0)
"""
# update capture time
VN_dispatcher.update_capture_time(capStop - capStart)
# Process image
# We schedule the process as opposed to waiting for an available core
# This brings consistancy and prevents overwriting a dead process before
# information has been grabbed from the Pipe
if VN_dispatcher.is_ready():
# queue the image for later use: displaying image, overlays, recording
imageQueue.append((frame, self.frames_captured))
# the function must be run directly from the class
VN_dispatcher.dispatch(target=detector.analyze_frame_async, args=(frame, self.frames_captured))
self.frames_captured += 1
# retreive results
if VN_dispatcher.is_available():
# results of image processor
results = VN_dispatcher.retreive()
# get image that was passed with the image processor
for f in imageQueue:
img, frame_id = f
if results[0] == frame_id:
imageQueue.remove(f)
break
VN_logger.text(VN_logger.GENERAL, "Frame {0}".format(frame_id))
# overlay gui
# rend_Image = gui.add_target_highlights(img, results[3])
rend_Image = gui.add_ring_highlights(img, results[4])
# show/record images
VN_logger.image(VN_logger.RAW, img)
VN_logger.image(VN_logger.GUI, rend_Image)
# display/log data
VN_logger.text(
VN_logger.ALGORITHM,
"RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}".format(
results[1], results[2], results[3], results[4]
),
)
VN_logger.text(VN_logger.DEBUG, "Rings detected: {0}".format(len(results[4])))
# send results if we found the landing pad
if results[2] is not None:
# shift origin to center of the image
x_pixel = results[2][0] - (self.camera_width / 2.0)
y_pixel = results[2][1] - (
self.camera_height / 2.0
) # y-axis is inverted??? Works with arducopter
# convert target location to angular radians
x_angle = x_pixel * (self.camera_hfov / self.camera_width) * (math.pi / 180.0)
y_angle = y_pixel * (self.camera_vfov / self.camera_height) * (math.pi / 180.0)
# send commands to autopilot
veh_control.report_landing_target(x_angle, y_angle, results[3], 0, 0)
else:
VN_logger.text(VN_logger.GENERAL, "Not in landing mode")
# terminate program
VN_logger.text(VN_logger.GENERAL, "Vehicle disconnected, Program Terminated")
if self.simulator == False:
VN_video.stop_capture()
readings += 1
#can not decode target
if(readings == 0):
return -1
#average all distance readings
return distance/(readings * 1.0)
if __name__ == "__main__":
cam = cv2.VideoCapture(0)
cam = flow_cam
detector = CircleDetector()
if cam is not None:
while True:
ret, img = cam.read()
if(img is not None and ret == True):
results = detector.analyze_frame(img)
rend_Image = gui.add_target_highlights(img, results[3])
#show results
cv2.imshow('gui', rend_Image)
cv2.waitKey(1)
print ('RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}'.format(results[0],results[1],results[2],results[3]))
else:
print "failed to grab image"
readings += 1
#can not decode target
if (readings == 0):
return -1
#average all distance readings
return distance / (readings * 1.0)
if __name__ == "__main__":
cam = cv2.VideoCapture(0)
cam = flow_cam
detector = CircleDetector()
if cam is not None:
while True:
ret, img = cam.read()
if (img is not None and ret == True):
results = detector.analyze_frame(img)
rend_Image = gui.add_target_highlights(img, results[3])
#show results
cv2.imshow('gui', rend_Image)
cv2.waitKey(1)
print('RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}'.
format(results[0], results[1], results[2], results[3]))
else:
print "failed to grab image"
#video source
cam = ImageReader(args.input)
file_count = len([f for f in os.walk(args.input).next()[2] if f[-4:] == ".bmp"])
key = np.empty((file_count),object)
print "total frames", file_count
frame_id = 0
if cam is not None and cam.isOpened():
while True:
ret, img = cam.peek(frame_id)
if(img is not None and ret == True):
img = gui.add_stats(img, str(frame_id) + '/' + str(file_count),10,10)
cv2.imshow('Image', img)
k = cv2.waitKey(0)
if k == 65362: #upkey
key[frame_id] = 1.0
frame_id += 1
print frame_id, "Present"
elif k == 65364: #down key
key[frame_id] = 0.0
frame_id += 1
print frame_id, "Not present"
elif k == 65363: #right key
key[frame_id] = 0.5
frame_id += 1
print frame_id, "Partial"
img *= (255.0 / (range_v))
img /= res_reduction
img *= res_reduction
return img
else:
return np.zeros((img.shape[0], img.shape[1]), np.uint8)
if __name__ == "__main__":
#cam = cv2.VideoCapture(0)
cam = flow_cam
detector = TopCode()
if cam is not None:
while True:
ret, img = cam.read()
if (img is not None and ret == True):
results = detector.analyze_frame(img)
print results[1]
rend_Image = gui.add_ring_highlights(img, results[3])
#show results
cv2.imshow('gui', rend_Image)
cv2.waitKey(1)
print('RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}'.
format(results[0], results[1], results[2], results[3]))
else:
print "failed to grab image"
def run(self):
VN_logger.text(VN_logger.GENERAL, 'Running {0}'.format(self.name()))
#start a video capture
if (self.simulator):
VN_logger.text(VN_logger.GENERAL, 'Using simulator')
sim.load_target(self.target_file, self.target_size)
sim.set_target_location(veh_control.get_home(True))
#sim.set_target_location(Location(0,0,0))
else:
VN_video.start_capture(self.camera_index)
#create an image processor
#detector = CircleDetector()
detector = TopCode()
#create a queue for images
imageQueue = []
#initilize autopilot variables
location = Location(0, 0, 0)
attitude = Attitude(0, 0, 0)
while veh_control.is_connected():
'''
#kill camera for testing
if(cv2.waitKey(2) == 1113938):
self.kill_camera = not self.kill_camera
'''
#we are in the landing zone or in a landing mode and we are still running the landing program
#just because the program is running does not mean it controls the vehicle
#i.e. in the landing area but not in a landing mode
#FIXME add inside_landing_area() back to conditional
if (self.always_run) or (veh_control.get_mode()
== "LAND") or (veh_control.get_mode()
== "RTL"):
#update how often we dispatch a command
VN_dispatcher.calculate_dispatch_schedule()
#update simulator
if (self.simulator):
#get info from autopilot
location = veh_control.get_location()
attitude = veh_control.get_attitude()
sim.refresh_simulator(location, attitude)
# grab an image
frame = None
capStart = current_milli_time()
if (self.simulator):
frame = sim.get_frame()
else:
frame = VN_video.get_image()
capStop = current_milli_time()
'''
if(self.kill_camera):
frame[:] = (0,255,0)
'''
#update capture time
VN_dispatcher.update_capture_time(capStop - capStart)
#Process image
#We schedule the process as opposed to waiting for an available core
#This brings consistancy and prevents overwriting a dead process before
#information has been grabbed from the Pipe
if VN_dispatcher.is_ready():
#queue the image for later use: displaying image, overlays, recording
imageQueue.append((frame, self.frames_captured))
#the function must be run directly from the class
VN_dispatcher.dispatch(target=detector.analyze_frame_async,
args=(frame, self.frames_captured))
self.frames_captured += 1
#retreive results
if VN_dispatcher.is_available():
#results of image processor
results = VN_dispatcher.retreive()
# get image that was passed with the image processor
for f in imageQueue:
img, frame_id = f
if results[0] == frame_id:
imageQueue.remove(f)
break
VN_logger.text(VN_logger.GENERAL,
'Frame {0}'.format(frame_id))
#overlay gui
#rend_Image = gui.add_target_highlights(img, results[3])
rend_Image = gui.add_ring_highlights(img, results[4])
#show/record images
VN_logger.image(VN_logger.RAW, img)
VN_logger.image(VN_logger.GUI, rend_Image)
#display/log data
VN_logger.text(
VN_logger.ALGORITHM,
'RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}'
.format(results[1], results[2], results[3],
results[4]))
VN_logger.text(
VN_logger.DEBUG,
'Rings detected: {0}'.format(len(results[4])))
#send results if we found the landing pad
if (results[2] is not None):
#shift origin to center of the image
x_pixel = results[2][0] - (self.camera_width / 2.0)
y_pixel = results[2][1] - (
self.camera_height / 2.0
) #y-axis is inverted??? Works with arducopter
#convert target location to angular radians
x_angle = x_pixel * (self.camera_hfov /
self.camera_width) * (math.pi /
180.0)
y_angle = y_pixel * (self.camera_vfov /
self.camera_height) * (math.pi /
180.0)
#send commands to autopilot
veh_control.report_landing_target(
x_angle, y_angle, results[3], 0, 0)
else:
VN_logger.text(VN_logger.GENERAL, 'Not in landing mode')
#terminate program
VN_logger.text(VN_logger.GENERAL,
'Vehicle disconnected, Program Terminated')
if (self.simulator == False):
VN_video.stop_capture()
img -= min_v
img *= (255.0/(range_v))
img /= res_reduction
img *= res_reduction
return img
else:
return np.zeros((img.shape[0],img.shape[1]), np.uint8)
if __name__ == "__main__":
#cam = cv2.VideoCapture(0)
cam = flow_cam
detector = TopCode()
if cam is not None:
while True:
ret, img = cam.read()
if(img is not None and ret == True):
results = detector.analyze_frame(img)
print results[1]
rend_Image = gui.add_ring_highlights(img, results[3])
#show results
cv2.imshow('gui', rend_Image)
cv2.waitKey(1)
print ('RunTime: {0} Center: {1} Distance: {2} Raw Target: {3}'.format(results[0],results[1],results[2],results[3]))
else:
print "failed to grab image"
#video source
cam = ImageReader(args.input)
file_count = len(
[f for f in os.walk(args.input).next()[2] if f[-4:] == ".bmp"])
key = np.empty((file_count), object)
print "total frames", file_count
frame_id = 0
if cam is not None and cam.isOpened():
while True:
ret, img = cam.peek(frame_id)
if (img is not None and ret == True):
img = gui.add_stats(img,
str(frame_id) + '/' + str(file_count), 10, 10)
cv2.imshow('Image', img)
k = cv2.waitKey(0)
if k == 65362: #upkey
key[frame_id] = 1.0
frame_id += 1
print frame_id, "Present"
elif k == 65364: #down key
key[frame_id] = 0.0
frame_id += 1
print frame_id, "Not present"
elif k == 65363: #right key
key[frame_id] = 0.5
frame_id += 1
print frame_id, "Partial"
