def add_user(img_path, group_name):
"""
Add a single user, save face in user image as feature vector.
:param img_path: Path to user image
:param group_name: Name of user group
:return: None
"""
db_path = os.path.join(DB_ROOT_DIR, group_name)
if not os.path.exists(db_path):
os.makedirs(db_path)
if not os.path.exists(img_path):
print("Invalid Image Path: " + img_path + "!")
return
extension = img_path.split(".")[-1]
if extension != "jpg" and extension != "png" and extension != 'jfif':
print("Invalid Image Path: " + img_path + "!")
return
img = imageio.imread(img_path)
prewhitened = detect_face(img)
if prewhitened is not None:
img_name = str(img_path.split("/")[-1].split(".")[0])
imageio.imsave(os.path.join(db_path, img_name + '.jpg'), prewhitened)
print(img_name + " Has Been Added!")
def add_user_batch(img_path, group_name):
"""
Add users in batches, read all images in the path and save faces as feature vectors
:param img_path: Path to the folder holding user images
:param group_name: Name of user group
:return: None
"""
db_path = os.path.join(DB_ROOT_DIR, group_name)
if not os.path.exists(db_path):
os.makedirs(db_path)
if not os.path.exists(img_path):
print("Invalid Image Path: " + img_path + "!")
return
for image in os.listdir(img_path):
extension = image.split(".")[-1]
if extension == "jpg" or extension == "png" or extension == "jfif":
img = imageio.imread(os.path.join(img_path, image))
prewhitened = detect_face(img)
if prewhitened is not None:
img_name = str(image.split('.')[0])
imageio.imsave(os.path.join(db_path, img_name + '.jpg'),
prewhitened)
print(str(image.split('.')[0]) + " Has Been Added!")
def find_faces(self, image):
faces = []
bounding_boxes, _ = detect_face(image, self.minsize, self.pnet,
self.rnet, self.onet, self.threshold,
self.factor)
for bb in bounding_boxes:
face = Face()
face.container_image = image
face.bounding_box = np.zeros(4, dtype=np.int32)
img_size = np.asarray(image.shape)[0:2]
face.bounding_box[0] = np.maximum(
bb[0] - self.face_crop_margin / 2, 0)
face.bounding_box[1] = np.maximum(
bb[1] - self.face_crop_margin / 2, 0)
face.bounding_box[2] = np.minimum(
bb[2] + self.face_crop_margin / 2, img_size[1])
face.bounding_box[3] = np.minimum(
bb[3] + self.face_crop_margin / 2, img_size[0])
cropped = image[face.bounding_box[1]:face.bounding_box[3],
face.bounding_box[0]:face.bounding_box[2], :]
face.image = misc.imresize(
cropped, (self.face_crop_size, self.face_crop_size),
interp='bilinear')
faces.append(face)
return faces
def read(self):
file_name, ok = QFileDialog.getOpenFileNames(self, '多文件选择', './')
if ok:
if len(file_name)>9:
local_path = os.path.dirname(os.path.dirname(file_name[0]))
if os.path.exists(local_path+"/detect_face")!=0 :
if os.path.exists( local_path+"/embedding.txt")==0:
embedding_image(local_path,self.images_placeholder,self.embeddings,self.phase_train_placeholder,self.sess)
train_model(local_path)
if os.path.exists(local_path+"/detect_face")==0:
detect_face(local_path,file_name)
embedding_image(local_path,self.images_placeholder,self.embeddings,self.phase_train_placeholder,self.sess)
#for roots,dirs,files in os.walk("./new_model"):
#print(dirs)
#print(os.path.dirname(file_name[0]))
train_model(os.path.dirname(file_name[0]))
QMessageBox.information(self, "Message", "success",QMessageBox.Ok, QMessageBox.Ok)
def run(input_image):
if (graph == None or model == None):
print("Run setup function once first")
return []
faces = detect_face(input_image)
if (faces == 0):
print("No face detected")
return []
face_input_image = faces[0]
infer_image = cv2.imread(face_input_image)
input_vector = utilities.run_inference(infer_image, graph)
match = utilities.run_image(model, input_vector)
os.remove(face_input_image)
return [match]
def add_user_batch(img_path, group_name):
"""
Add users in batches, read all images in the path and save faces as feature vectors
:param img_path: Path to the folder holding user images
:param group_name: Name of user group
:return: None
"""
db_path = os.path.join(DB_ROOT_DIR, group_name)
if not os.path.exists(db_path):
os.makedirs(db_path)
if not os.path.exists(img_path):
print("Invalid Image Path: " + img_path + "!")
return
for image in os.listdir(img_path):
extension = image.split(".")[-1]
if extension == "jpg" or extension == "png" or extension == "jfif":
img = cv2.cvtColor(cv2.imread(os.path.join(img_path, image)), cv2.COLOR_BGR2RGB)
face_locations = detect_face(img)
if len(face_locations) > 1:
print("More than One Face in Image: " + image + "!")
continue
if len(face_locations) < 1:
print("No Faces in Image: " + image + "!")
continue
# Encode faces into 128-dimensional features
face_enc = face_recognition.face_encodings(face_image=img,
known_face_locations=face_locations,
num_jitters=1
)[0]
name = str(image.split(".")[0])
np.save(os.path.join(db_path, name), face_enc)
print(name + " Has Been Added!")
print("User Add Finished!")
def add_user(img_path, group_name):
"""
Add a single user, save face in user image as feature vector.
:param img_path: Path to user image
:param group_name: Name of user group
:return: None
"""
db_path = os.path.join(DB_ROOT_DIR, group_name)
if not os.path.exists(db_path):
os.makedirs(db_path)
if not os.path.exists(img_path):
print("Invalid Image Path: " + img_path + "!")
return
extension = img_path.split(".")[-1]
if extension != "jpg" and extension != "png" and extension != "jfif":
print("Invalid Image Path: " + img_path + "!")
return
img = cv2.cvtColor(cv2.imread(img_path), cv2.COLOR_BGR2RGB)
face_locations = detect_face(img)
if len(face_locations) > 1:
print("More than One Face in Image!")
return
if len(face_locations) < 1:
print("No Faces in Image!")
return
# Encode faces into 128-dimensional features
face_enc = face_recognition.face_encodings(face_image=img,
known_face_locations=face_locations,
num_jitters=1
)[0]
name = str(img_path.split(".")[-2].split("/")[-1])
np.save(os.path.join(db_path, name), face_enc)
print(name + " Has Been Added!")
def main():
detect_face(1)
def realtime_recognition(group_name,
face_size=1,
track_interval=200,
recognition_interval=2000,
scale_factor=1,
tolerance=0.6):
"""
Run realtime face recognition.
:param group_name: Name of user group
:param face_size: Minimum detected face size
:param track_interval: Face detect interval/ms
:param recognition_interval: Face recognize interval/ms
:param scale_factor: Image processing zoom factor
:param tolerance: Face recognition threshold
:return: None
"""
# Load face database
db_path = os.path.join(DB_ROOT_DIR, group_name)
known_face_encodings = [] # Features in database
known_face_names = [] # Names in database
for person in os.listdir(db_path):
known_face_encodings.append(np.load(os.path.join(db_path, person)))
known_face_names.append(person.replace(".", "_").split("_")[0])
face_locations = [] # Container for detected face boxes
face_names = [] # Container for recognized face names
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1920)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 1080)
ret, frame = cap.read()
timer = 0 # Frame skip timer
while ret:
timer += 1
ret, frame = cap.read()
# Face detection
if timer % (track_interval * FPS // 1000) == 0:
small_frame = cv2.resize(frame, (0, 0),
fx=1 / scale_factor,
fy=1 / scale_factor)
rgb_small_frame = cv2.cvtColor(small_frame, cv2.COLOR_BGR2RGB)
# gray_small_frame = cv2.cvtColor(small_frame, cv2.COLOR_BGR2GRAY)
face_locations = detect_face(rgb_small_frame, face_size)
# Face recognition
if timer % (recognition_interval * FPS //
1000) == 0 and face_locations != []:
rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# Encode faces into 128-dimensional features
face_encodings = face_recognition.face_encodings(
face_image=rgb_frame,
known_face_locations=face_locations * scale_factor,
num_jitters=1)
face_names.clear()
for face_encoding in face_encodings:
matches = face_recognition.compare_faces(
known_face_encodings=known_face_encodings,
face_encoding_to_check=face_encoding,
tolerance=tolerance)
name = "Unknown"
face_distances = face_recognition.face_distance(
face_encodings=known_face_encodings,
face_to_compare=face_encoding)
best_match_index = np.argmin(face_distances)
if matches[best_match_index]:
name = known_face_names[int(best_match_index)]
face_names.append(name)
# Draw face boxes and names
for (top, right, bottom, left), name in zip(face_locations,
face_names):
top *= scale_factor
right *= scale_factor
bottom *= scale_factor
left *= scale_factor
cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 1)
cv2.rectangle(frame, (left, bottom),
(right, int(bottom + (bottom - top) * 0.25)),
(0, 0, 255), cv2.FILLED)
cv2.putText(frame, name,
(left, int(bottom + (bottom - top) * 0.24)),
cv2.FONT_HERSHEY_DUPLEX, (right - left) / 120,
(255, 255, 255), 1)
cv2.imshow('camera', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyWindow("camera")
def align_mtcnn(input_dir,
output_dir,
image_size=182,
margin=44,
random_order=None,
gpu_memory_fraction=1.0,
detect_multiple_faces=False):
"""
Align dataset
:param input_dir:
:param output_dir:
:param image_size:
:param margin:
:param random_order:
:param gpu_memory_fraction:
:param detect_multiple_faces:
:return:
"""
sleep(random.random())
output_dir = os.path.expanduser(output_dir)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
# Store some git revision info in a text file in the log directory
src_path, _ = os.path.split(os.path.realpath(__file__))
store_revision_info(src_path, output_dir, ' '.join(sys.argv))
dataset = get_dataset(input_dir)
print('Creating networks and loading parameters')
with tf.Graph().as_default():
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=gpu_memory_fraction)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options, log_device_placement=False))
with sess.as_default():
pnet, rnet, onet = create_mtcnn(sess, None)
minsize = 20 # minimum size of face
threshold = [0.6, 0.7, 0.7] # three steps's threshold
factor = 0.709 # scale factor
# Add a random key to the filename to allow alignment using multiple processes
random_key = np.random.randint(0, high=99999)
bounding_boxes_filename = os.path.join(output_dir, 'bounding_boxes_%05d.txt' % random_key)
with open(bounding_boxes_filename, "w") as text_file:
nrof_images_total = 0
nrof_successfully_aligned = 0
if random_order:
random.shuffle(dataset)
for cls in dataset:
output_class_dir = os.path.join(output_dir, cls.name)
if not os.path.exists(output_class_dir):
os.makedirs(output_class_dir)
if random_order:
random.shuffle(cls.image_paths)
for image_path in cls.image_paths:
nrof_images_total += 1
filename = os.path.splitext(os.path.split(image_path)[1])[0]
output_filename = os.path.join(output_class_dir, filename + '.png')
print(image_path)
if not os.path.exists(output_filename):
try:
img = misc.imread(image_path)
except (IOError, ValueError, IndexError) as e:
errorMessage = '{}: {}'.format(image_path, e)
print(errorMessage)
else:
if img.ndim < 2:
print('Unable to align "%s"' % image_path)
text_file.write('%s\n' % (output_filename))
continue
if img.ndim == 2:
img = to_rgb(img)
img = img[:, :, 0:3]
bounding_boxes, _ = detect_face(img, minsize, pnet, rnet, onet, threshold, factor)
nrof_faces = bounding_boxes.shape[0]
if nrof_faces > 0:
det = bounding_boxes[:, 0:4]
det_arr = []
img_size = np.asarray(img.shape)[0:2]
if nrof_faces > 1:
if detect_multiple_faces:
for i in range(nrof_faces):
det_arr.append(np.squeeze(det[i]))
else:
bounding_box_size = (det[:, 2] - det[:, 0]) * (det[:, 3] - det[:, 1])
img_center = img_size / 2
offsets = np.vstack([(det[:, 0] + det[:, 2]) / 2 - img_center[1],
(det[:, 1] + det[:, 3]) / 2 - img_center[0]])
offset_dist_squared = np.sum(np.power(offsets, 2.0), 0)
index = np.argmax(
bounding_box_size - offset_dist_squared * 2.0) # some extra weight on the centering
det_arr.append(det[index, :])
else:
det_arr.append(np.squeeze(det))
for i, det in enumerate(det_arr):
det = np.squeeze(det)
bb = np.zeros(4, dtype=np.int32)
bb[0] = np.maximum(det[0] - margin / 2, 0)
bb[1] = np.maximum(det[1] - margin / 2, 0)
bb[2] = np.minimum(det[2] + margin / 2, img_size[1])
bb[3] = np.minimum(det[3] + margin / 2, img_size[0])
cropped = img[bb[1]:bb[3], bb[0]:bb[2], :]
scaled = misc.imresize(cropped, (image_size, image_size), interp='bilinear')
nrof_successfully_aligned += 1
filename_base, file_extension = os.path.splitext(output_filename)
if detect_multiple_faces:
output_filename_n = "{}_{}{}".format(filename_base, i, file_extension)
else:
output_filename_n = "{}{}".format(filename_base, file_extension)
misc.imsave(output_filename_n, scaled)
text_file.write('%s %d %d %d %d\n' % (output_filename_n, bb[0], bb[1], bb[2], bb[3]))
else:
print('Unable to align "%s"' % image_path)
text_file.write('%s\n' % (output_filename))
print('Total number of images: %d' % nrof_images_total)
print('Number of successfully aligned images: %d' % nrof_successfully_aligned)
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--path',
help='Path of the video you want to test on.',
default=0)
parser.add_argument('--image-size', default='112,112', help='')
parser.add_argument('--image', default='Tom_Hanks_54745.png', help='')
parser.add_argument('--model',
default='../gender-age/model/model,0',
help='path to load model.')
parser.add_argument('--gpu', default=0, type=int, help='gpu id')
parser.add_argument(
'--det',
default=0,
type=int,
help='mtcnn option, 1 means using R+O, 0 means detect from begining')
args = parser.parse_args()
MINSIZE = 20
THRESHOLD = [0.6, 0.7, 0.7]
FACTOR = 0.709
IMAGE_SIZE = 182
INPUT_IMAGE_SIZE = 160
CLASSIFIER_PATH = 'Models/Entity/Entity_margin_svm.pkl'
VIDEO_PATH = args.path
FACENET_MODEL_PATH = 'Models/facenet/20180402-114759.pb'
# Load The Custom Classifier
with open(CLASSIFIER_PATH, 'rb') as file:
model, class_names = pickle.load(file)
print("Custom Classifier, Successfully loaded")
# # Load age and gender model
# depth = 16
# k = 8
# margin = 0.4
# weight_file = "Models/weights.28-3.73.hdf5"
# model_age_gender = WideResNet(64, depth=depth, k=k)()
# model_age_gender.load_weights(weight_file)
model_age_gender = face_model.FaceModel(args)
with tf.Graph().as_default():
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.6)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options,
log_device_placement=False))
with sess.as_default():
# Load the model
print('Loading feature extraction model')
facenet.load_model(FACENET_MODEL_PATH)
# Get input and output tensors
images_placeholder = tf.get_default_graph().get_tensor_by_name(
"input:0")
embeddings = tf.get_default_graph().get_tensor_by_name(
"embeddings:0")
phase_train_placeholder = tf.get_default_graph(
).get_tensor_by_name("phase_train:0")
embedding_size = embeddings.get_shape()[1]
pnet, rnet, onet = align.detect_face.create_mtcnn(
sess, "./src/align")
people_detected = set()
person_detected = collections.Counter()
frame = cv2.imread(VIDEO_PATH)
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# bounding_boxes, _ = align.detect_face.detect_face(frame, MINSIZE, pnet, rnet, onet, THRESHOLD, FACTOR)
# bounding_boxes = detect_tiny_face(frame)
bounding_boxes = detect_face(frame)
print(bounding_boxes)
print(len(bounding_boxes))
print(bounding_boxes.shape)
faces_found = bounding_boxes.shape[0]
entity = {}
try:
if faces_found > 0:
det = bounding_boxes[:, 0:4]
bb = np.zeros((faces_found, 4), dtype=np.int32)
unknown_faces = []
unknown_coor = []
for i in range(faces_found):
bb[i][0] = det[i][0]
bb[i][1] = det[i][1]
bb[i][2] = det[i][2]
bb[i][3] = det[i][3]
x, y = bb[i][0], bb[i][1]
w = bb[i][2] - x
h = bb[i][3] - y
x_pad_size = round(w * 0.01)
y_pad_size = round(h * 0.01)
x -= x_pad_size
y -= y_pad_size
w += 2 * x_pad_size
h += 2 * y_pad_size
bb[i][0] = x
bb[i][1] = y
bb[i][2] = x + w
bb[i][3] = y + h
cropped = frame[bb[i][1]:bb[i][3],
bb[i][0]:bb[i][2], :]
cropped_ = frame[bb[i][1]:bb[i][3], bb[i][0]:bb[i][2]]
scaled = cv2.resize(
cropped, (INPUT_IMAGE_SIZE, INPUT_IMAGE_SIZE),
interpolation=cv2.INTER_CUBIC)
scaled = facenet.prewhiten(scaled)
scaled_reshape = scaled.reshape(
-1, INPUT_IMAGE_SIZE, INPUT_IMAGE_SIZE, 3)
feed_dict = {
images_placeholder: scaled_reshape,
phase_train_placeholder: False
}
emb_array = sess.run(embeddings, feed_dict=feed_dict)
predictions = model.predict_proba(emb_array)
best_class_indices = np.argmax(predictions, axis=1)
best_class_probabilities = predictions[
np.arange(len(best_class_indices)),
best_class_indices]
best_name = class_names[best_class_indices[0]]
print("Name: {}, Probability: {}".format(
best_name, best_class_probabilities))
color = (0, 0, 0)
if best_class_probabilities > 0.75:
name = class_names[best_class_indices[0]]
if name == "unknown":
name_entity = "{}_{}".format(name, i)
unknown_faces.append(
cv2.resize(np.copy(cropped_), (64, 64),
interpolation=cv2.INTER_CUBIC))
unknown_coor.append(
(bb[i][0], bb[i][1], bb[i][2], bb[i][3]))
# color = (255, 255, 255)
aligned = np.transpose(cropped_, (2, 0, 1))
input_blob = np.expand_dims(aligned, axis=0)
data = mx.nd.array(input_blob)
db = mx.io.DataBatch(data=(data, ))
gender, age = model_age_gender.get_ga(db)
print(gender, age)
coor = (bb[i][0], bb[i][1], bb[i][2], bb[i][3])
entity[coor] = "{}_{}".format(
"F" if gender == 0 else "M", age)
else:
name_entity = name
coor = (bb[i][0], bb[i][1], bb[i][2], bb[i][3])
entity[coor] = name_entity
# color = (0, 0, 255)
# cv2.rectangle(frame, (bb[i][0], bb[i][1]), (bb[i][2], bb[i][3]), color, 2)
# text_x = bb[i][0]
# text_y = bb[i][3] + 20
# cv2.putText(frame, name, (text_x, text_y), cv2.FONT_HERSHEY_COMPLEX_SMALL,
# 1, color, thickness=1, lineType=2)
# cv2.putText(frame, str(round(best_class_probabilities[0], 3)), (text_x, text_y+17), cv2.FONT_HERSHEY_COMPLEX_SMALL,
# 1, color, thickness=1, lineType=2)
person_detected[best_name] += 1
else:
unknown_faces.append(
cv2.resize(np.copy(cropped_), (64, 64),
interpolation=cv2.INTER_CUBIC))
unknown_coor.append(
(bb[i][0], bb[i][1], bb[i][2], bb[i][3]))
# cv2.rectangle(frame, (bb[i][0], bb[i][1]), (bb[i][2], bb[i][3]), (255, 255, 255), 2)
# text_x = bb[i][0]
# text_y = bb[i][3] + 20
name = "unknown"
aligned = np.transpose(cropped_, (2, 0, 1))
input_blob = np.expand_dims(aligned, axis=0)
data = mx.nd.array(input_blob)
db = mx.io.DataBatch(data=(data, ))
gender, age = model_age_gender.get_ga(db)
print(gender, age)
coor = (bb[i][0], bb[i][1], bb[i][2], bb[i][3])
entity[coor] = "{}_{}".format(
"F" if gender == 0 else "M", age)
# name_entity = "{}_{}".format(name,i)
# cv2.putText(frame, name, (text_x, text_y), cv2.FONT_HERSHEY_COMPLEX_SMALL,
# 1, (255, 255, 255), thickness=1, lineType=2)
# cv2.putText(frame, str(round(best_class_probabilities[0], 3)), (text_x, text_y+17), cv2.FONT_HERSHEY_COMPLEX_SMALL,
# 1, (255, 255, 255), thickness=1, lineType=2)
person_detected[best_name] += 1
# entity[name_entity] = [bb[i][1], bb[i][3], bb[i][0], bb[i][2]]
except Exception as e:
print(e)
pass
# unknown_faces_stack = np.stack(unknown_faces, axis=0)
# print(unknown_faces_stack.shape)
# print(type(unknown_faces_stack))
# results = model_age_gender.predict(unknown_faces_stack)
# print("hahah")
# predicted_genders = results[0]
# ages = np.arange(0, 101).reshape(101, 1)
# predicted_ages = results[1].dot(ages).flatten()
# print(predicted_genders)
# print(predicted_ages)
# for i in range(len(predicted_ages)):
# label = "{}_{}".format(int(predicted_ages[i]), "M" if predicted_genders[i][0] < 0.5 else "F")
# entity[unknown_coor[i]] = label
print(entity)
for key, value in entity.items():
if value.split("_")[-1] == "M" or value.split("_")[-1] == "F":
color = (255, 255, 255)
else:
color = (0, 0, 255)
text_x = key[0]
text_y = key[3] + 20
cv2.rectangle(frame, (key[0], key[1]), (key[2], key[3]), color, 2)
cv2.putText(frame,
value, (text_x, text_y),
cv2.FONT_HERSHEY_COMPLEX_SMALL,
1, (255, 255, 255),
thickness=1,
lineType=2)
cv2.imshow('Face Recognition', frame)
cv2.waitKey(0)
