def predict_image(self, image):
"""Keras predict gender and age"""
img_data = image[np.newaxis, :]
img_data = preprocess_input(img_data)
results = self.model.predict(img_data)
return results
def extract_feature(dir_path, net):
features = []
person_id = []
infos = []
cam_id = []
for image_name in sorted(os.listdir(dir_path)):
if '.txt' in image_name:
continue
if 'f' in image_name or 's' in image_name:
arr = image_name.split('_')
person = int(arr[0])
camera = int(arr[1][1])
elif 's' not in image_name:
# grid
print(image_name)
arr = image_name.split('_')
person = int(arr[0])
camera = int(arr[1])
else:
continue
image_path = os.path.join(dir_path, image_name)
img = image.load_img(image_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
feature = net.predict(x)
features.append(np.squeeze(feature))
person_id.append(person)
cam_id.append(camera)
infos.append((person, camera))
return features, infos, person_id, cam_id
def load_data(data_name, path):
data_loader = DataManager(data_name,
image_size=input_shape[:2],
dataset_path=path)
faces, emotions = data_loader.get_data()
faces = preprocess_input(faces)
num_samples, num_classes = emotions.shape
image_size = faces.shape[1]
train_data, val_data = split_data(faces, emotions, validation_split)
return train_data, val_data, image_size, num_classes
def predict_gender(name):
#loading data and images
image_path = source + name
m = re.search('.*(?=-)', name)
if m:
found = m.group(0)
else:
found = name
result = {'male': 0, 'female': 0, 'domain': found}
try:
rgb_image = load_image(image_path, grayscale=False)
except:
print('3. Doesn"t open')
if (os.path.isfile(image_path)):
os.remove(image_path)
to_remove.append(name)
return result
gray_image = load_image(image_path, grayscale=True)
gray_image = np.squeeze(gray_image)
gray_image = gray_image.astype('uint8')
#face and gender detection
faces = detect_faces(face_detection, gray_image)
if (len(faces) == 0):
print('no faces')
to_remove.append(name)
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, gender_offsets)
rgb_face = rgb_image[y1:y2, x1:x2]
try:
rgb_face = cv2.resize(rgb_face, (gender_target_size))
except:
continue
rgb_face = preprocess_input(rgb_face, False)
rgb_face = np.expand_dims(rgb_face, 0)
gender_prediction = gender_classifier.predict(rgb_face)
gender_label_arg = np.argmax(gender_prediction)
gender_text = gender_labels[gender_label_arg]
if gender_text == gender_labels[0]:
result['female'] += 1
else:
result['male'] += 1
return result
#bgr_image = np.empty((480 * 640 * 3,), dtype=np.uint8)
#camera.capture(bgr_image, format='bgr', use_video_port=True)
bgr_image = bgr_image.reshape((480, 640, 3))
gray_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2GRAY)
rgb_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2RGB)
faces = detect_faces(face_detection, gray_image)
for face_coordinates in faces:
count = count + 1
x1, x2, y1, y2 = apply_offsets(face_coordinates, emotion_offsets)
gray_face = gray_image[y1:y2, x1:x2]
try:
gray_face = cv2.resize(gray_face, (emotion_target_size))
except:
continue
gray_face = preprocess_input(gray_face, True) # 类似于标准化的操作(64,64)
gray_face = np.expand_dims(gray_face, 0) # (1,64,64)
gray_face = np.expand_dims(gray_face, -1) # (1,64,64,1)
emotion_prediction = emotion_classifier.predict(gray_face)
emotion_probability = np.max(emotion_prediction)
emotion_label_arg = np.argmax(emotion_prediction)
emotion_text = emotion_labels[emotion_label_arg]
emotion_window.append(emotion_text)
if len(emotion_window) > frame_window:
emotion_window.pop(0)
try:
emotion_mode = mode(emotion_window)
except:
continue
def preprocess_images(self, image_array):
return preprocess_input(image_array)
# callback
# hdf5 형식으로 저장
log_file_path = base_path + dataset_name + '_emotion_training.log'
csv_logger = CSVLogger(log_file_path, append=False)
early_stop = EarlyStopping('val_loss', patience=patience)
reduce_lr = ReduceLROnPlateau('val_loss',
factor=0.1,
patience=int(patience / 4),
verbose=1)
trained_models_path = base_path = dataset_name + '_mini_Xception'
model_names = trained_models_path + '.{epoch:02d}-{val_loss:.2f}.hdf5'
model_checkpoint = ModelCheckpoint(model_names,
'val_loss',
verbose=1,
save_best_only=True)
callbacks = [model_checkpoint, csv_logger, early_stop, reduce_lr]
#loading dataset
data_loader = DataManager(dataset_name, image_size=input_shape[:2])
faces, emotions = data_loader.get_data()
faces = preprocess_input(faces)
train_data, val_data = split_data(faces, emotions, validation_split)
train_faces, train_emotions = train_data
model.fit_generator(data_generator.flow(train_faces, train_emotions,
batch_size),
steps_per_epoch=len(train_faces) / batch_size,
epochs=num_epochs,
verbose=1,
callbacks=callbacks,
validation_data=val_data)
def preprocess(image_arr):
data = preprocess_input(image_arr)
return data
def crop_face(file_name, face_detection, name_count):
face_detection_size = (40, 40)
counter = 0
frame_process_counter = 0
# starting video streaming
cv2.namedWindow('Attendence_Tracker', cv2.WINDOW_NORMAL)
# cv2.namedWindow('Attendence_Tracker')
# file_name = '../top10/person1.mp4'
video_capture = cv2.VideoCapture(file_name)
time.sleep(1.0)
while (video_capture.isOpened()):
ret, bgr_image = video_capture.read()
if ret == False:
break
counter += 1
if counter % 1 == 0:
frame_process_counter += 1
gray_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2GRAY)
rgb_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2RGB)
faces = detect_faces(face_detection, bgr_image,confidence)
count = 0
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, face_offsets)
rgb_face = rgb_image[y1:y2, x1:x2]
print("len", len(rgb_face))
# print(rgb_face)
if len(rgb_face) != 0 and counter % 1 ==0:
print(detector.detect_faces(rgb_face))
dict_mtcnn = detector.detect_faces(rgb_face)
if len(dict_mtcnn) != 0:
bounding_box = dict_mtcnn[0]['box']
new_image = rgb_image[bounding_box[2]:bounding_box[3], bounding_box[0]:bounding_box[1]]
cv2.rectangle(new_image,
(bounding_box[0], bounding_box[1]),
(bounding_box[0]+bounding_box[2], bounding_box[1] + bounding_box[3]),
(0,155,255), 2)
# cv2.imwrite("align/align_{}/align_{}_{}".format(name_count, name_count,counter) + ".jpg", cv2.cvtColor(img, cv2.COLOR_RGB2BGR))
cv2.imwrite("align/emp_{}/emp_{}_{}".format(name_count, name_count,counter) + ".jpg", cv2.cvtColor(rgb_face, cv2.COLOR_RGB2BGR))
print("image saved-------------------", counter)
count += 1
try:
rgb_face = cv2.resize(rgb_face, (face_detection_size))
except:
continue
rgb_face = np.expand_dims(rgb_face, 0)
rgb_face = preprocess_input(rgb_face, False)
# Bounding box color
color = (255, 0, 0)
identity = "this is me"
draw_bounding_box(face_coordinates, rgb_image, color, identity)
bgr_image = cv2.cvtColor(rgb_image, cv2.COLOR_RGB2BGR)
cv2.imshow('Attendence_Tracker', bgr_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
print('Total frames processed:', counter, frame_process_counter)
break
video_capture.release()
# out.release()
cv2.destroyAllWindows()
return "successful"
faces = face_cascade.detectMultiScale(gray_image,
scaleFactor=1.1,
minNeighbors=5,
minSize=(30, 30),
flags=cv2.CASCADE_SCALE_IMAGE)
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, emotion_offsets)
gray_face = gray_image[y1:y2, x1:x2]
try:
gray_face = cv2.resize(gray_face, (emotion_target_size))
except:
continue
gray_face = preprocess_input(gray_face, True)
gray_face = np.expand_dims(gray_face, 0)
gray_face = np.expand_dims(gray_face, -1)
emotion_prediction = emotion_classifier.predict(gray_face)
emotion_probability = np.max(emotion_prediction)
emotion_label_arg = np.argmax(emotion_prediction)
emotion_text = emotion_labels[emotion_label_arg]
emotion_window.append(emotion_text)
if len(emotion_window) > frame_window:
emotion_window.pop(0)
try:
emotion_mode = mode(emotion_window)
except:
continue
start = time.clock()
face_cascade = cv2.CascadeClassifier(
'/home/user/anaconda3/pkgs/libopencv-3.4.2-hb342d67_1/share/OpenCV/haarcascades/haarcascade_frontalface_default.xml'
)
faces = face_cascade.detectMultiScale(gray_image, 1.3, 5)
print(np.shape(faces))
# faces = detect_faces(face_detection, gray_image)
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, emotion_offsets)
rgb_face = rgb_image[y1:y2, x1:x2]
gray_face = gray_image[y1:y2, x1:x2]
try:
if not is_light_net:
face = cv2.resize(rgb_face, (target_size),
interpolation=cv2.INTER_CUBIC)
face = preprocess_input(face, False)
face = np.expand_dims(face, 0)
else:
face = cv2.resize(gray_face, (target_size),
interpolation=cv2.INTER_CUBIC)
face = preprocess_input(face, False)
face = np.expand_dims(face, 0)
face = np.expand_dims(face, -1)
except:
continue
start_time = time.clock()
if is_single_task:
emotion_label = emotion_model.predict(face)
gender_label = gender_model.predict(face)
pose_label = pose_model.predict(face)
inner_padding_factor = 0.1
outer_padding = 0
output_size = 160
reference_5pts = get_reference_facial_points(
(output_size, output_size), inner_padding_factor,
(outer_padding, outer_padding), default_square)
dst_img = warp_and_crop_face(raw,
facial5points,
reference_pts=reference_5pts,
crop_size=(crop_size, crop_size))
# (3) Convert image data to keras format
img_data = dst_img[np.newaxis, :]
img_data = preprocess_input(img_data)
data_results = model.predict(img_data)
t1 = time.time() - t0
print("Time used: {} ms".format(t1 * 1000))
# (4) Predict gender and other attributes
predicted_gender = gCategory[np.argmax(data_results[0])]
ages = np.arange(0, 71).reshape(71, 1)
predicted_age = int(np.round(data_results[1].dot(ages).flatten()[0],
1))
# (5) Draw images
if predicted_gender == 'W':
cv.rectangle(raw, (box[0], box[1]), (box[2], box[3]), (0, 0, 255),
2)
else:
def emotion():
# parameters for loading data and images
emotion_text = ""
detection_model_path = '../trained_models/detection_models/haarcascade_frontalface_default.xml'
emotion_model_path = '../trained_models/emotion_models/fer2013_mini_XCEPTION.102-0.66.hdf5'
emotion_labels = get_labels('fer2013')
# hyper-parameters for bounding boxes shape
frame_window = 10
emotion_offsets = (20, 40)
# loading models
face_detection = load_detection_model(detection_model_path)
emotion_classifier = load_model(emotion_model_path, compile=False)
# getting input model shapes for inference
emotion_target_size = emotion_classifier.input_shape[1:3]
# starting lists for calculating modes
emotion_window = []
# starting video streaming
cv2.namedWindow('window_frame')
video_capture = cv2.VideoCapture(0)
while True:
bgr_image = video_capture.read()[1]
gray_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2GRAY)
rgb_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2RGB)
faces = detect_faces(face_detection, gray_image)
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, emotion_offsets)
gray_face = gray_image[y1:y2, x1:x2]
try:
gray_face = cv2.resize(gray_face, (emotion_target_size))
except:
continue
gray_face = preprocess_input(gray_face, True)
gray_face = np.expand_dims(gray_face, 0)
gray_face = np.expand_dims(gray_face, -1)
emotion_prediction = emotion_classifier.predict(gray_face)
emotion_probability = np.max(emotion_prediction)
emotion_label_arg = np.argmax(emotion_prediction)
emotion_text = emotion_labels[emotion_label_arg]
emotion_window.append(emotion_text)
if len(emotion_window) > frame_window:
emotion_window.pop(0)
try:
emotion_mode = mode(emotion_window)
except:
continue
if emotion_text == 'angry':
color = emotion_probability * np.asarray((255, 0, 0))
elif emotion_text == 'sad':
color = emotion_probability * np.asarray((0, 0, 255))
elif emotion_text == 'happy':
color = emotion_probability * np.asarray((255, 255, 0))
elif emotion_text == 'surprise':
color = emotion_probability * np.asarray((0, 255, 255))
else:
color = emotion_probability * np.asarray((0, 255, 0))
color = color.astype(int)
color = color.tolist()
draw_bounding_box(face_coordinates, rgb_image, color)
draw_text(face_coordinates, rgb_image, emotion_mode, color, 0, -45,
1, 1)
if (emotion_text):
return emotion_text
break
bgr_image = cv2.cvtColor(rgb_image, cv2.COLOR_RGB2BGR)
cv2.imshow('window_frame', bgr_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
def load_image(image_array):
image_array = np.expand_dims(image_array, axis=0)
image_array = preprocess_input(image_array)
return image_array
def crop_face(file_name, face_detection, dirName):
dire = "cropped_faces/" + dirName
try:
os.makedirs(dire)
print("Directory ", dire, " Created ")
except FileExistsError:
print("Directory ", dire, " already exists")
face_detection_size = (40, 40)
counter = 0
frame_process_counter = 0
# starting video streaming
cv2.namedWindow('Attendence_Tracker', cv2.WINDOW_NORMAL)
# cv2.namedWindow('Attendence_Tracker')
# file_name = '../top10/person1.mp4'
video_capture = cv2.VideoCapture(file_name)
time.sleep(1.0)
while (video_capture.isOpened()):
ret, bgr_image = video_capture.read()
if ret == False:
break
counter += 1
if counter % 1 == 0:
frame_process_counter += 1
gray_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2GRAY)
rgb_image = cv2.cvtColor(bgr_image, cv2.COLOR_BGR2RGB)
faces = detect_faces(face_detection, bgr_image, confidence)
count = 0
for face_coordinates in faces:
x1, x2, y1, y2 = apply_offsets(face_coordinates, face_offsets)
rgb_face = rgb_image[y1:y2, x1:x2]
print("len", len(rgb_face))
# print(rgb_face)
if len(rgb_face) != 0 and counter % 1 == 0:
cv2.imwrite(
dire + "/" + dirName + "_{}".format(counter) + ".jpg",
cv2.cvtColor(rgb_face, cv2.COLOR_RGB2BGR))
print("image saved-------------------", counter)
count += 1
try:
rgb_face = cv2.resize(rgb_face, (face_detection_size))
except:
continue
rgb_face = np.expand_dims(rgb_face, 0)
rgb_face = preprocess_input(rgb_face, False)
# Bounding box color
color = (255, 0, 0)
identity = "this is me"
draw_bounding_box(face_coordinates, rgb_image, color, identity)
bgr_image = cv2.cvtColor(rgb_image, cv2.COLOR_RGB2BGR)
cv2.imshow('Attendence_Tracker', bgr_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
print('Total frames processed:', counter, frame_process_counter)
break
video_capture.release()
# out.release()
cv2.destroyAllWindows()
return "successful"
