def __init__(self, is_site):
# load classifier
if (is_site):
self.model = SiteModel(
"light_classification/models/frozen_inference_graph.pb")
else:
self.model = SimModel()
def run():
from sim_model import SimModel
from sim_array import SimArray
pstudy = SimArray(sim_model=SimModel())
pstudy_view = SimArrayView(model=pstudy)
pstudy_view.configure_traits()
def __init__(self, scenario):
if scenario == "sim":
self.model = SimModel()
else:
# self.model = RealModel("light_classification/models/tl_model")
self.model = RealModel(
"light_classification/models/frozen_inference_graph.pb")
def __init__(self, scenario):
if scenario == "sim":
self.model = SimModel()
else:
# self.model = RealModel("light_classification/models/tl_model")
self.model = RealModel(
"light_classification/frozen_models/faster_rcnn_resnet101_coco_2018_01_28/frozen_inference_graph.pb"
)
def __init__(self, is_sim):
self.is_sim = is_sim
if is_sim:
self.model = SimModel()
else:
self.threshold = THRESHOLD
self.graph = self.load_graph(GRAPH_PATH_SIM)
self.image_tensor = self.graph.get_tensor_by_name('image_tensor:0')
self.detect_boxes = self.graph.get_tensor_by_name(
'detection_boxes:0')
self.detect_scores = self.graph.get_tensor_by_name(
'detection_scores:0')
self.detect_classes = self.graph.get_tensor_by_name(
'detection_classes:0')
self.num_detections = self.graph.get_tensor_by_name(
'num_detections:0')
self.sess = tf.Session(graph=self.graph)
def run():
from sim_model import SimModel
sim_array = SimArray(sim_model=SimModel())
print sim_array.levels2run[:, 0, 0, 0]
print sim_array[:, 0, 0, 0]
print sim_array.run_table[0]
print 'array_content', sim_array.output_array[0, 0, 0, 0, :]
sim_array.clear_cache()
print 'array_content', sim_array.output_array[0, 0, 0, 0, :]
sim_array.configure_traits()
class TLClassifier(object):
def __init__(self, is_site):
# load classifier
if (is_site):
self.model = SiteModel(
"light_classification/models/frozen_inference_graph.pb")
else:
self.model = SimModel()
def get_classification(self, image):
"""Determines the color of the traffic light in the image
Args:
image (cv::Mat): image containing the traffic light
Returns:
int: ID of traffic light color (specified in styx_msgs/TrafficLight)
"""
# implement light color prediction
return self.model.predict(image)
# A trait is regarded as Factor if it specifies
# a set of levels in form ps_levels = True.
# - get_sim_outputs() as instances of SimOut class
# defining the name and order of outputs
# - peval() method returning a vector of results
# the order of outputs is prescribed by the
# specification given in the get_outputs() method
#
# Here we just import predefined Foo model with three four inputs
#
# [ index_1, material_model, param_1, param_2 ]
#
# The inputs have the types [ Int, Callable, Float, Float
#
from sim_model import SimModel
sim_model = SimModel()
# The model response is obtained by issuing
#
print 'default evaluation', sim_model.peval()
# returning an array with two values.
# In this call, the default values of the factors
# [input_1, material_model, param_1, param_2 ]
# were taken. The factor levels were ignored.
# DEFINING A STUDY
# --------------
# In order to study the response of the model in a broader range
# of specified levels we now construct the parametric study
#
def run():
from sim_model import SimModel
pstudy_app = SimPStudy(sim_model=SimModel())
pstudy_app.configure_traits(kind='live')
class TLClassifier(object):
def __init__(self, is_sim):
self.is_sim = is_sim
if is_sim:
self.model = SimModel()
else:
self.threshold = THRESHOLD
self.graph = self.load_graph(GRAPH_PATH_SIM)
self.image_tensor = self.graph.get_tensor_by_name('image_tensor:0')
self.detect_boxes = self.graph.get_tensor_by_name(
'detection_boxes:0')
self.detect_scores = self.graph.get_tensor_by_name(
'detection_scores:0')
self.detect_classes = self.graph.get_tensor_by_name(
'detection_classes:0')
self.num_detections = self.graph.get_tensor_by_name(
'num_detections:0')
self.sess = tf.Session(graph=self.graph)
def get_classification(self, image):
"""Determines the color of the traffic light in the image
Args:
image (cv::Mat): image containing the traffic light, BGR channel
Returns:
int: ID of traffic light color (specified in styx_msgs/TrafficLight)
"""
if self.is_sim:
return self.model.predict(image)
else:
with self.graph.as_default():
image_expanded = np.expand_dims(image, axis=0)
(boxes, scores, classes, num) = self.sess.run(
[
self.detect_boxes, self.detect_scores,
self.detect_classes, self.num_detections
],
feed_dict={self.image_tensor: image_expanded})
boxes = np.squeeze(boxes)
scores = np.squeeze(scores)
classes = np.squeeze(classes)
if scores[0] > self.threshold:
if classes[0] == 1:
return TrafficLight.GREEN
elif classes[0] == 2:
return TrafficLight.RED
elif classes[0] == 3:
return TrafficLight.YELLOW
return TrafficLight.UNKNOWN
def load_graph(self, graph_path):
graph = tf.Graph()
with graph.as_default():
graph_def = tf.GraphDef()
with tf.gfile.GFile(graph_path, 'rb') as fid:
read_graph = fid.read()
graph_def.ParseFromString(read_graph)
tf.import_graph_def(graph_def, name='')
return graph
def _sim_model_default(self):
from sim_model import SimModel
return SimModel()
# A trait is regarded as Factor if it specifies
# a set of levels in form ps_levels = True.
# - get_sim_outputs() as instances of SimOut class
# defining the name and order of outputs
# - peval() method returning a vector of results
# the order of outputs is prescribed by the
# specification given in the get_outputs() method
#
# Here we just import predefined Foo model with three four inputs
#
# [ index_1, material_model, param_1, param_2 ]
#
# The inputs have the types [ Int, Callable, Float, Float
#
from sim_model import SimModel
sim_model = SimModel()
# The model response is obtained by issueing
#
print 'default evaluation', sim_model.peval()
# returning an array with two values.
# In this call, the default values of the factors
# [input_1, material_model, param_1, param_2 ]
# were taken. The factor levels were ignored.
# DEFINING A STUDY
# --------------
# In order to study the response of the model in a broader range
# of specified levels we now construct the parametric study
#