def load_models(self):
""" Loads all models. """
face_detector_net = self.load_model(
self.Helpers.confs["genisysai"]["detection"])
face_detector_net.reshape({"data": [1, 3, 384, 672]})
landmarks_net = self.load_model(
self.Helpers.confs["genisysai"]["landmarks"])
face_reid_net = self.load_model(
self.Helpers.confs["genisysai"]["reidentification"])
self.face_detector = FaceDetector(face_detector_net,
confidence_threshold=0.6,
roi_scale_factor=1.15)
self.landmarks_detector = LandmarksDetector(landmarks_net)
self.face_identifier = FaceIdentifier(face_reid_net,
match_threshold=0.3,
match_algo='HUNGARIAN')
self.face_detector.deploy(self.Helpers.confs["genisysai"]["runas"], self.context)
self.landmarks_detector.deploy(self.Helpers.confs["genisysai"]["runas"], self.context,
queue_size=self.qs)
self.face_identifier.deploy(self.Helpers.confs["genisysai"]["runas"], self.context,
queue_size=self.qs)
self.Helpers.logger.info("Models loaded")
class GeniSysAI():
def __init__(self):
""" GeniSysAI Class
GeniSysAI functions for the Hospital Intelligent Automation System.
"""
self.Helpers = Helpers("GeniSysAI", False)
self.qs = 16
self.context = InferenceContext([self.Helpers.confs["genisysai"]["runas"], self.Helpers.confs["genisysai"]["runas"], self.Helpers.confs["genisysai"]["runas"]], "", "", "")
self.Helpers.logger.info("GeniSysAI Helper Class initialization complete.")
def connect(self):
""" Connects to the local GeniSysAI camera. """
self.lcv = cv2.VideoCapture(self.Helpers.confs["genisysai"]["vid"])
self.Helpers.logger.info("Connected to GeniSysAI")
def load_models(self):
""" Loads all models. """
face_detector_net = self.load_model(
self.Helpers.confs["genisysai"]["detection"])
face_detector_net.reshape({"data": [1, 3, 384, 672]})
landmarks_net = self.load_model(
self.Helpers.confs["genisysai"]["landmarks"])
face_reid_net = self.load_model(
self.Helpers.confs["genisysai"]["reidentification"])
self.face_detector = FaceDetector(face_detector_net,
confidence_threshold=0.6,
roi_scale_factor=1.15)
self.landmarks_detector = LandmarksDetector(landmarks_net)
self.face_identifier = FaceIdentifier(face_reid_net,
match_threshold=0.3,
match_algo='HUNGARIAN')
self.face_detector.deploy(self.Helpers.confs["genisysai"]["runas"], self.context)
self.landmarks_detector.deploy(self.Helpers.confs["genisysai"]["runas"], self.context,
queue_size=self.qs)
self.face_identifier.deploy(self.Helpers.confs["genisysai"]["runas"], self.context,
queue_size=self.qs)
self.Helpers.logger.info("Models loaded")
def load_model(self, model_path):
""" Loads a model from path. """
model_path = osp.abspath(model_path)
model_weights_path = osp.splitext(model_path)[0] + ".bin"
self.Helpers.logger.info("Loading the model from '%s'" % (model_path))
model = self.context.ie_core.read_network(model_path, model_weights_path)
self.Helpers.logger.info("Model loaded")
return model
def load_known(self):
""" Loads known data. """
self.faces_database = FacesDatabase(self.Helpers.confs["genisysai"]["data"], self.face_identifier,
self.landmarks_detector, self.face_detector, True)
self.face_identifier.set_faces_database(self.faces_database)
self.Helpers.logger.info("Database is built, registered %s identities" %
(len(self.faces_database)))
def process(self, frame):
""" Processes a frame. """
orig_image = frame.copy()
frame = frame.transpose((2, 0, 1))
frame = np.expand_dims(frame, axis=0)
self.face_detector.clear()
self.landmarks_detector.clear()
self.face_identifier.clear()
self.face_detector.start_async(frame)
rois = self.face_detector.get_roi_proposals(frame)
if self.qs < len(rois):
self.Helpers.logger.info("Too many faces for processing." \
" Will be processed only %s of %s." % \
(self.qs, len(rois)))
rois = rois[:self.qs]
self.landmarks_detector.start_async(frame, rois)
landmarks = self.landmarks_detector.get_landmarks()
self.face_identifier.start_async(frame, rois, landmarks)
face_identities, unknowns = self.face_identifier.get_matches()
outputs = [rois, landmarks, face_identities]
return outputs
def draw_text_with_background(self, frame, text, origin,
font=cv2.FONT_HERSHEY_SIMPLEX, scale=1.0,
color=(0, 0, 0), thickness=1, bgcolor=(255, 255, 255)):
text_size, baseline = cv2.getTextSize(text, font, scale, thickness)
cv2.rectangle(frame,
tuple((origin + (0, baseline)).astype(int)),
tuple((origin + (text_size[0], -text_size[1])).astype(int)),
bgcolor, cv2.FILLED)
cv2.putText(frame, text,
tuple(origin.astype(int)),
font, scale, color, thickness)
return text_size, baseline
def draw_detection_roi(self, frame, roi, identity):
label = self.face_identifier.get_identity_label(identity.id)
# Draw face ROI border
cv2.rectangle(frame,
tuple(roi.position), tuple(roi.position + roi.size),
(0, 220, 0), 2)
# Draw identity label
text_scale = 0.5
font = cv2.FONT_HERSHEY_SIMPLEX
text_size = cv2.getTextSize("H1", font, text_scale, 1)
line_height = np.array([0, text_size[0][1]])
if label is "Unknown":
text = label
else:
text = "User #" + label
if identity.id != FaceIdentifier.UNKNOWN_ID:
text += ' %.2f%%' % (100.0 * (1 - identity.distance))
self.draw_text_with_background(frame, text,
roi.position - line_height * 0.5,
font, scale=text_scale)
return frame, label
def draw_detection_keypoints(self, frame, roi, landmarks):
keypoints = [landmarks.left_eye,
landmarks.right_eye,
landmarks.nose_tip,
landmarks.left_lip_corner,
landmarks.right_lip_corner,
landmarks.right_lip_corner]
for point in keypoints:
center = roi.position + roi.size * point
cv2.circle(frame, tuple(center.astype(int)), 2, (0, 255, 255), 2)
return frame
def __init__(self,
path,
face_identifier,
landmarks_detector,
face_detector=None,
no_show=False):
path = osp.abspath(path)
self.fg_path = path
self.no_show = no_show
paths = []
if osp.isdir(path):
paths = [osp.join(path, f) for f in os.listdir(path) \
if f.split('.')[-1] in self.IMAGE_EXTENSIONS]
else:
log.error("Wrong face images database path. Expected a " \
"path to the directory containing %s files, " \
"but got '%s'" % \
(" or ".join(self.IMAGE_EXTENSIONS), path))
if len(paths) == 0:
log.error("The images database folder has no images.")
self.database = []
for num, path in enumerate(paths):
label = osp.splitext(osp.basename(path))[0]
image = cv2.imread(path, flags=cv2.IMREAD_COLOR)
assert len(image.shape) == 3, \
"Expected an input image in (H, W, C) format"
assert image.shape[2] in [3, 4], \
"Expected BGR or BGRA input"
orig_image = image.copy()
image = image.transpose((2, 0, 1)) # HWC to CHW
image = np.expand_dims(image, axis=0)
if face_detector:
face_detector.start_async(image)
rois = face_detector.get_roi_proposals(image)
if len(rois) < 1:
log.warning("Not found faces on the image '%s'" % (path))
else:
w, h = image.shape[-1], image.shape[-2]
rois = [FaceDetector.Result([0, 0, 0, 0, 0, w, h])]
for i, roi in enumerate(rois):
r = [roi]
landmarks_detector.start_async(image, r)
landmarks = landmarks_detector.get_landmarks()
face_identifier.start_async(image, r, landmarks)
descriptor = face_identifier.get_descriptors()[0]
if face_detector:
mm = self.check_if_face_exist(
descriptor, face_identifier.get_threshold())
if mm < 0:
crop = orig_image[int(roi.position[1]):int(roi.position[1]+roi.size[1]), \
int(roi.position[0]):int(roi.position[0]+roi.size[0])]
name = self.ask_to_save(crop)
if name is None:
name = label
self.dump_faces(crop, descriptor, name)
else:
log.debug("Adding label {} to the gallery.".format(label))
self.add_item(descriptor, label)