def realsense():
stream = request.get_json()
#iterate over each element in json
f = Faces()
faces = []
if stream != None:
for i in stream:
if i is not None:
if 'people' in i and stream[i] is not None:
for j in range(0,len(stream[i])):
p = Face()
p.id = stream[i][j]['ID']
p.point.x = (stream[i][j]['z'])/1000.0
p.point.y = -(stream[i][j]['x'])/1000.0 # Center of the face
p.point.z = (stream[i][j]['y'])/1000.0
# p.pose.x = stream[i][j]['rx']
# p.pose.y = stream[i][j]['ry']
# p.pose.z = stream[i][j]['rz']
# p.pose.w = stream[i][j]['rw']
p.attention = stream[i][j]['confidence'] # This is an indication that the depth image is near or far a way. Low values indicate that it's not a good value.
#Now let's get the emotion related data out of the elements.
# if stream[i][j]['expression'] is not None:
# for expressions in j['expression']: # Now this equivalent makes sure that there is alway the expression being processed in the realsense demo application
# p.emotions.append(str(expressions))
# p.emotion_values.append(stream[i][j]['expression'][expressions])
faces.append(p)
if len(faces) > 0:
f.faces = faces # Check if it's not empty before publishing it.
facePub.publish(f)
return Response(json_encode({'success': True}), mimetype="application/json")
def publish_faces(self):
faces =[]
for f in self.faces.keys():
if self.faces[f].is_trackable():
face = Face()
face.id = f
face.point = self.faces[f].get_filtered_3d_point()
faces.append(face)
msg = Faces()
msg.faces = faces
self.faces_pub.publish(msg)
def look_at():
performance = request.get_data()
js = json.loads(performance)
# js = {'x':1,'y':0.3,'z':-0.5}
f = Face()
f.id = 1
f.point.x = js["x"]
f.point.y = -js["y"]
f.point.z = js["z"]
f.attention = 0.99
facePub.publish(Faces([f]))
return json_encode({'result': True})
def realsense():
stream = request.get_json()
#iterate over each element in json
f = Faces()
faces = []
if stream != None:
for i in stream:
if i is not None:
if 'people' in i and stream[i] is not None:
for j in range(0, len(stream[i])):
p = Face()
p.id = stream[i][j]['ID']
p.point.x = (stream[i][j]['z']) / 1000.0
p.point.y = -(
stream[i][j]['x']) / 1000.0 # Center of the face
p.point.z = (stream[i][j]['y']) / 1000.0
# p.pose.x = stream[i][j]['rx']
# p.pose.y = stream[i][j]['ry']
# p.pose.z = stream[i][j]['rz']
# p.pose.w = stream[i][j]['rw']
p.attention = stream[i][j][
'confidence'] # This is an indication that the depth image is near or far a way. Low values indicate that it's not a good value.
#Now let's get the emotion related data out of the elements.
# if stream[i][j]['expression'] is not None:
# for expressions in j['expression']: # Now this equivalent makes sure that there is alway the expression being processed in the realsense demo application
# p.emotions.append(str(expressions))
# p.emotion_values.append(stream[i][j]['expression'][expressions])
faces.append(p)
if len(faces) > 0:
f.faces = faces # Check if it's not empty before publishing it.
facePub.publish(f)
return Response(json_encode({'success': True}),
mimetype="application/json")
def callback(data):
global previous_faces
current_faces = {};
for user in data.tf_bundles:
user_id = int(re.sub("user_","",user.user_id))
location = user.head.translation
current_faces[user_id] = location
new_face = findDiff(list(current_faces.keys()), list(previous_faces.keys()))
lost_face = findDiff(list(previous_faces.keys()), list(current_faces.keys()))
# Publishers
# if new face is found
if (new_face):
face_event_msg = FaceEvent()
face_event_msg.face_event = "new_face"
face_event_msg.face_id = new_face
event_publisher.publish(face_event_msg)
# if face is lost
if (lost_face):
face_event_msg = FaceEvent()
face_event_msg.face_event = "lost_face"
face_event_msg.face_id = lost_face
event_publisher.publish(face_event_msg)
# Face Location
faces_msg = Faces()
faces_msg.faces = []
for user, location in current_faces.items():
face_msg = Face()
face_msg.id = user
face_msg.point = location
face_msg.attention = 1
faces_msg.faces.append(face_msg)
face_publisher.publish(faces_msg)
previous_faces = current_faces
