def __init__(
self,
image_size,
learning_rate=2e-5,
batch_size=1,
classes_size=2,
ngf=64,
):
"""
Args:
input_size：list [H, W, C]
batch_size: integer, batch size
learning_rate: float, initial learning rate for Adam
ngf: number of gen filters in first conv layer
"""
self.learning_rate = learning_rate
self.input_shape = [
int(batch_size / 4), image_size[0], image_size[1], image_size[2]
]
self.tenaor_name = {}
self.classes_size = classes_size
self.G_X = Unet('G_X',
ngf=ngf,
output_channl=image_size[2],
keep_prob=0.97)
self.D_X = Discriminator('D_X', ngf=ngf, keep_prob=0.9)
self.G_L_X = Detector('G_L_X',
ngf,
classes_size=classes_size,
keep_prob=0.99,
input_channl=image_size[2])
def __init__(self):
# params
self.count = 1
self.clear_str = '0:0: '
self.is_configuring = False
self.is_self_driving = False
self.is_time_configuring = False
self.first_load = True
self.engine = None
self.video_recorder = None
self.path = rospack.get_path('team107') + '/scripts/'
self.bt1_status = False
self.bt2_status = False
self.bt3_status = False
self.bt4_status = False
self.ss_status = False
self.model = Model(self.path)
self.sign_model = Detector(self.path)
# ros subscribers and publishers
# hal
self.sub_bt1 = rospy.Subscriber('/bt1_status',
Bool,
self.bt1_callback,
queue_size=1)
self.sub_bt2 = rospy.Subscriber('/bt2_status',
Bool,
self.bt2_callback,
queue_size=1)
self.sub_bt3 = rospy.Subscriber('/bt3_status',
Bool,
self.bt3_callback,
queue_size=1)
self.sub_bt4 = rospy.Subscriber('/bt4_status',
Bool,
self.bt4_callback,
queue_size=1)
self.sub_ss = rospy.Subscriber('/ss_status',
Bool,
self.ss_callback,
queue_size=1)
self.pub_lcd = rospy.Publisher('/lcd_print', String, queue_size='5')
# car controller
self.sub_img = None
rospy.init_node('team107')
print 'Finish initializing'
def __init__(self):
self.detector = Detector()
@route('/object', method='POST')
def object():
try:
image = request.files.get('file')
image = io.imread(StringIO(image.file.read()))
objs = detector(image.astype(np.float32) / 255.)
return json.dumps(objs)
except Exception as e:
print str(type(e)), e
@route('/style', method='POST')
def style():
try:
image = request.files.get('file')
image = io.imread(StringIO(image.file.read()))
result_image = style(image)
s = StringIO()
np.save(s, result_image)
return s.getvalue()
except Exception as e:
print str(type(e)), e
if __name__ == '__main__':
detector = Detector()
style = Style()
run(host='0.0.0.0', port=8080, debug=True, reloader=True)