def run(self):
while not rospy.is_shutdown():
msg = CNN_out()
msg.header.stamp = rospy.Time.now()
data = None
while data is None:
try:
data = rospy.wait_for_message("camera", Image, timeout=10)
except:
pass
self.use_network_out = True
# delete in performance
if self.use_network_out:
print("Publishing commands!")
else:
print("NOT Publishing commands!")
cv_image = utils.callback_img(data, self.target_size,
self.crop_size, self.imgs_rootpath,
self.use_network_out)
outs = self.model.predict_on_batch(cv_image[None])
steer, coll = outs[0][0], outs[1][0]
msg.steering_angle = steer
msg.collision_prob = coll
self.pub.publish(msg)
def run(self):
t = time.time()
while not rospy.is_shutdown():
msg = CNN_out()
msg.header.stamp = rospy.Time.now()
data = None
rate = rospy.Rate(60)
while self.rdy is False:
rate.sleep()
self.rdy = False
if self.use_network_out:
print("Publishing commands!")
#else:
# print("NOT Publishing commands!")
cv_image = utils.callback_img(self.data, self.target_size,
self.crop_size, self.imgs_rootpath,
self.use_network_out)
#print("###\n###\n")
#print("$%6.3f, $%6.3f, $%6.3f, $%6.3f" % (cv_image[0,0,0], cv_image[0,20,0], cv_image[0,30,0], cv_image[0,50,0]))
#print("###\n###\n")
outs = self.model.predict_on_batch(cv_image[None])
steer, coll = outs[0][0], outs[1][0]
print("{0}: got commands steer={1}, coll={2}".format(
time.time() - t, outs[0][0], outs[1][0]))
#toc = time.time()
#if ( (toc - t) >= 1 ):
# t = toc
# print("1 sec mark")
msg.steering_angle = steer
msg.collision_prob = coll
self.pub.publish(msg)
def run(self):
while not rospy.is_shutdown():
msg = CNN_out()
msg.header.stamp = rospy.Time.now()
data = None
while data is None:
try:
data = rospy.wait_for_message("camera", Image, timeout=10)
except:
pass
cv_image = utils.callback_img(data, self.target_size,
self.imgs_rootpath)
outs = self.model.predict_on_batch(cv_image[None])
outs = outs * 48 + 48 # undo the normalization
x1, y1, x2, y2, x3, y3, x4, y4 = outs[0][0], outs[0][1], outs[0][
2], outs[0][3], outs[0][4], outs[0][5], outs[0][6], outs[0][7]
# four corners
msg.left_up_x = x1
msg.left_up_y = y1
msg.right_up_x = x2
msg.right_up_y = y2
msg.right_down_x = x3
msg.right_down_y = y3
msg.left_down_x = x4
msg.left_down_y = y4
# calculate the waypoints ahead and after the window
self.pub.publish(msg) # publish eight numbers
def run(self):
while not rospy.is_shutdown():
msg = CNN_out()
msg.header.stamp = rospy.Time.now()
data = None
while data is None:
try:
data = rospy.wait_for_message(
"/bebop2/camera_base/image_raw", Image, timeout=10)
except:
pass
if self.use_network_out:
print("Publishing commands!")
else:
print("NOT Publishing commands!")
# Note that this call will save images if self.use_network_out is True (when DroNet takes over)
# cv_image = utils.callback_img(data, self.target_size, self.crop_size,
# self.imgs_rootpath, self.use_network_out)
cv_image = utils.callback_img(data, self.target_size,
self.crop_size, self.imgs_rootpath,
False)
outs = self.model.predict_on_batch(cv_image[None])
steer, coll = outs[0][0], outs[1][0]
msg.steering_angle = steer
msg.collision_prob = coll
print("Steering: {}\nCollision: {}".format(steer, coll))
self.pub.publish(msg)
def run(self):
while not rospy.is_shutdown():
msg = CNN_out()
msg.header.stamp = rospy.Time.now()
data = None
while data is None:
try:
data = rospy.wait_for_message("camera", Image, timeout=10)
except:
pass
if self.use_network_out:
print("Publishing commands!")
else:
print("NOT Publishing commands!")
cv_image = utils.callback_img(data, self.target_size,
self.crop_size, self.imgs_rootpath,
self.use_network_out)
outs = self.model.predict_on_batch(cv_image[None])
if self.get_saliency_maps:
img, cam_s, cam_c = utils.grad_cam(self.model,
cv_image[None],
self.crop_size,
layer_name="add_3",
scale_factor=2)
cam_s_msg = utils.np2image(cam_s)
cam_c_msg = utils.np2image(cam_c)
img_msg = utils.np2image(img)
self.gradcam_nn_pub.publish(img_msg)
self.gradcam_col_pub.publish(cam_c_msg)
self.gradcam_st_pub.publish(cam_s_msg)
else:
img = np.array(cv_image * 255, dtype=np.uint8)
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
img_msg = utils.np2image(img)
self.gradcam_nn_pub.publish(img_msg)
steer, coll = outs[0][0], outs[1][0]
msg.steering_angle = steer * 30
msg.collision_prob = coll
self.pub.publish(msg)