def train_recognition(model_detector, model_encoder, model_classifier,
face_embeddings_path, verify):
ensure_directory(config.INPUT_DIR_DATASET)
print("")
names = get_dataset_names(config.INPUT_DIR_DATASET)
if names is not None:
print("Names " + str(names))
for name in names:
for (_d, _n,
files) in os.walk(config.INPUT_DIR_DATASET + "/" + name):
print(name + ": " + str(files))
print("")
ensure_directory(config.INPUT_DIR_MODEL_TRAINING)
face_detector = FaceDetector(model=model_detector,
path=config.INPUT_DIR_MODEL_DETECTION)
face_encoder = FaceEncoder(model=model_encoder,
path=config.INPUT_DIR_MODEL_ENCODING,
path_training=config.INPUT_DIR_MODEL_TRAINING,
training=True)
face_encoder.train(face_detector,
path_dataset=face_embeddings_path,
verify=verify,
classifier=model_classifier)
def process_facerecognition():
cam_index = 0
cam_resolution = RESOLUTION_QVGA
model_detector = FaceDetectorModels.HAARCASCADE
# model_detector=FaceDetectorModels.DLIBHOG
# model_detector=FaceDetectorModels.DLIBCNN
# model_detector=FaceDetectorModels.SSDRESNET
# model_detector=FaceDetectorModels.MTCNN
model_recognizer = FaceEncoderModels.LBPH
# model_recognizer=FaceEncoderModels.OPENFACE
# model_recognizer=FaceEncoderModels.DLIBRESNET
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
while (True):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Display updated frame to web app
yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' +
cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
# Release the camera
camera.release()
cv2.destroyAllWindows()
def process_facerecognition(model_detector, model_recognizer, cam_index,
cam_resolution):
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
while (True):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
break
# Display updated frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
# Release the camera
camera.release()
cv2.destroyAllWindows()
def train_recognition(model_detector, model_encoder, model_classifier, verify):
ensure_directory(INPUT_DIR_DATASET)
ensure_directory(INPUT_DIR_MODEL_TRAINING)
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
face_encoder = FaceEncoder(model=model_encoder,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=True)
face_encoder.train(face_detector,
path_dataset=INPUT_DIR_DATASET,
verify=verify,
classifier=model_classifier)
def process_facerecognition(model_detector, model_recognizer, image):
# Initialize the camera
image = cv2.VideoCapture(image)
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
try:
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except Exception as ex:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
print(ex)
face_id, confidence = (None, 0)
# Capture frame-by-frame
ret, frame = image.read()
if ret == 0:
print("Unexpected error! " + image)
return
# frame = cv2.imread(image)
# Detect faces in the image
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(frame, (x, y, w, h))
print(face_id)
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Display the resulting frame
cv2.imshow(WINDOW_NAME, frame)
cv2.waitKey(1)
# Release the image
image.release()
cv2.destroyAllWindows()
def init_model(int_detect=1, int_encode=3):
# Only for debugging while developing
try:
# Initialize face detection
global face_recognizer, face_detector, face_encoder
face_detector = FaceDetector(model=FaceDetectorModels(int_detect),
path=join(ROOT_DIR,
INPUT_DIR_MODEL_DETECTION))
# Initialize face recognizer
face_encoder = FaceEncoder(model=FaceEncoderModels(int_encode),
path=join(ROOT_DIR,
INPUT_DIR_MODEL_ENCODING),
path_training=join(
ROOT_DIR, INPUT_DIR_MODEL_TRAINING),
training=False)
face_recognizer = FaceRecognizer(face_embeddings_path=join(
ROOT_DIR, EMBEDDINGS_DIR + '/3_face_encodings_83.pickle'))
except Exception as ex:
face_encoder = None
print(ex)
def process_livenessdetection(model_detector, model_recognizer, model_liveness,
cam_index, cam_resolution):
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
##input
check = int(
input(
"Enter number 1 if you want to show live video and vice versa: "))
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize face liveness detection
face_liveness = FaceLiveness(
model=FaceLivenessModels.EYESBLINK_MOUTHOPEN,
path=INPUT_DIR_MODEL_LIVENESS)
face_liveness2 = FaceLiveness(
model=FaceLivenessModels.COLORSPACE_YCRCBLUV,
path=INPUT_DIR_MODEL_LIVENESS)
except:
print("Error, check if models and trained dataset models exists!")
return
face_id, confidence = (None, 0)
eyes_close, eyes_ratio = (False, 0)
total_eye_blinks, eye_counter, eye_continuous_close = (
0, 0, 1) # eye_continuous_close should depend on frame rate
mouth_open, mouth_ratio = (False, 0)
total_mouth_opens, mouth_counter, mouth_continuous_open = (
0, 0, 1) # eye_continuous_close should depend on frame rate
time_start = time()
time_elapsed = 0
frame_count = 0
identified_unique_faces = {} # dictionary
runtime = 60 # monitor for 10 seconds only
is_fake_count_print = 0
is_fake_count_replay = 0
face_count = 0
##edit
time_recognition = 5
checkface = False
print("Note: this will run for {} seconds only".format(runtime))
while (time_elapsed < runtime):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
# Check if eyes are close and if mouth is open
eyes_close, eyes_ratio = face_liveness.is_eyes_close(frame, face)
mouth_open, mouth_ratio = face_liveness.is_mouth_open(frame, face)
#print("eyes_close={}, eyes_ratio ={:.2f}".format(mouth_open, mouth_ratio))
#print("mouth_open={}, mouth_ratio={:.2f}".format(mouth_open, mouth_ratio))
# Detect if frame is a print attack or replay attack based on colorspace
is_fake_print = face_liveness2.is_fake(frame, face)
is_fake_replay = face_liveness2.is_fake(frame, face, flag=1)
# Identify face only if it is not fake and eyes are open and mouth is close
if is_fake_print:
is_fake_count_print += 1
face_id, confidence = ("Fake", None)
elif is_fake_replay:
is_fake_count_replay += 1
face_id, confidence = ("Fake", None)
elif not eyes_close and not mouth_open:
face_id, confidence = face_encoder.identify(frame, face)
if (face_id
not in identified_unique_faces) & (confidence > 50):
identified_unique_faces[face_id] = 1
elif (face_id in identified_unique_faces) & (confidence > 50):
identified_unique_faces[face_id] += 1
if (face_count > 100) | (face_id == "Fake"):
face_count = 0
elif (face_id != "Fake") & (confidence > 50):
face_count += 1
print("Identifying: {:.2f} %".format((face_count / 31) * 100))
label_face(frame, face, face_id,
confidence) # Set text and bounding box on face
break # Process 1 face only
# Monitor eye blinking and mouth opening for liveness detection
total_eye_blinks, eye_counter = monitor_eye_blinking(
eyes_close, eyes_ratio, total_eye_blinks, eye_counter,
eye_continuous_close)
total_mouth_opens, mouth_counter = monitor_mouth_opening(
mouth_open, mouth_ratio, total_mouth_opens, mouth_counter,
mouth_continuous_open)
# Update frame count
frame_count += 1
time_elapsed = time() - time_start
# Display updated frame
if check == 1:
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
if face_count > 30:
checkface = True
break
else:
checkface = False
print("Note: this will run for {} seconds only".format(runtime))
# Determining if face is alive can depend on the following factors and more:
time_elapsed = int(time() - time_start)
print("\n")
print("Face Liveness Data:")
print("time_elapsed = {}".format(time_elapsed)
) # recognition will run for specific time (ex. 3 seconds)
print("frame_count = {}".format(
frame_count)) # can be used for averaging
print("total_eye_blinks = {}".format(
total_eye_blinks)) # fake face if 0
print("total_mouth_opens = {}".format(
total_mouth_opens)) # fake face if 0
print("is_fake_count_print = {}".format(
is_fake_count_print)) # fake face if not 0
print("is_fake_count_replay = {}".format(
is_fake_count_replay)) # fake face if not 0
print("identified_unique_faces = {}".format(
identified_unique_faces)) # fake face if recognized more than 1 face
print("Todo: determine if face is alive using this data.")
print("\n")
if checkface == True:
print("Hello {}!!!".format(
max(identified_unique_faces, key=identified_unique_faces.get)))
else:
print("Can not indentify your face, please try again!")
#print("{}".format(face_id))
# Release the camera
camera.release()
cv2.destroyAllWindows()
def process_facerecognition(model_detector, model_recognizer,
model_speech_synthesizer, model_speech_recognizer,
cam_index, cam_resolution):
# Initialize speech-to-text (speech recognizer) for voice-activated capability (wake-word/hot-word/trigger-word detection)
# Then wait for trigger word before starting face recognition
if True:
speech_recognizer = SpeechRecognizer(model=model_speech_recognizer,
path=None)
print("\nWaiting for a trigger word: {}".format(TRIGGER_WORDS))
speech_recognizer.start(TRIGGER_WORDS, speech_recognizer_callback)
global trigger_word_detected
try:
while (trigger_word_detected == False):
time.sleep(1)
except:
pass
speech_recognizer.stop()
speech_recognizer = None
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize text-to-speech (speech synthesizer) for voice-enabled capability
speech_synthesizer = SpeechSynthesizer(model=model_speech_synthesizer,
path=None,
path_output=None,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
# Start face recognition
frame_count = 0
while (True):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Play audio file corresponding to the recognized name
if (frame_count % 30 == 0):
if len(faces) == 1 and (face_id is not None) and (face_id !=
"Unknown"):
speech_synthesizer.playaudio(INPUT_DIR_AUDIOSET,
face_id,
block=False)
# Process 1 face only
break
# Display updated frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
frame_count += 1
# Release the camera
camera.release()
cv2.destroyAllWindows()
def process_facerecognition_livenessdetection_poseagegenderemotion(
cam_resolution,
out_resolution,
framecount,
image=None,
model_detector=0,
model_recognizer=0):
from libfaceid.liveness import FaceLivenessDetectorModels, FaceLiveness
from libfaceid.pose import FacePoseEstimatorModels, FacePoseEstimator
from libfaceid.age import FaceAgeEstimatorModels, FaceAgeEstimator
from libfaceid.gender import FaceGenderEstimatorModels, FaceGenderEstimator
from libfaceid.emotion import FaceEmotionEstimatorModels, FaceEmotionEstimator
model_poseestimator = FacePoseEstimatorModels.DEFAULT
model_ageestimator = FaceAgeEstimatorModels.DEFAULT
model_genderestimator = FaceGenderEstimatorModels.DEFAULT
model_emotionestimator = FaceEmotionEstimatorModels.DEFAULT
# Initialize the camera
if image is not None:
cap = cv2.VideoCapture(image)
else:
cap = cam_init(cam_resolution[0], cam_resolution[1])
###############################################################################
# FACE DETECTION
###############################################################################
# Initialize face detection
face_detector = FaceDetector(
model=model_detector,
path=INPUT_DIR_MODEL_DETECTION) #, optimize=True)
###############################################################################
# FACE RECOGNITION
###############################################################################
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
###############################################################################
# EYE BLINKING DETECTOR
###############################################################################
# Initialize detector for blinking eyes
face_liveness = FaceLiveness(model=FaceLivenessDetectorModels.EYEBLINKING,
path=INPUT_DIR_MODEL_ESTIMATION)
face_liveness.initialize()
(eye_counter, total_eye_blinks) = (0, 0)
###############################################################################
# FACE POSE/AGE/GENDER/EMOTION ESTIMATION
###############################################################################
# Initialize pose/age/gender/emotion estimation
if model_poseestimator is not None:
face_pose_estimator = FacePoseEstimator(
model=model_poseestimator, path=INPUT_DIR_MODEL_ESTIMATION)
if model_ageestimator is not None:
face_age_estimator = FaceAgeEstimator(model=model_ageestimator,
path=INPUT_DIR_MODEL_ESTIMATION)
if model_genderestimator is not None:
face_gender_estimator = FaceGenderEstimator(
model=model_genderestimator, path=INPUT_DIR_MODEL_ESTIMATION)
if model_emotionestimator is not None:
face_emotion_estimator = FaceEmotionEstimator(
model=model_emotionestimator, path=INPUT_DIR_MODEL_ESTIMATION)
(age, gender, emotion) = (None, None, None)
# Initialize fps counter
fps_frames = 0
fps_start = time()
fps = 0
saveVideo = False
out = None
while (True):
# Capture frame-by-frame
ret, frame = cap.read()
if ret == 0:
print("Unexpected error! " + image)
break
###############################################################################
# FACE DETECTION and FACE RECOGNITION
###############################################################################
# Detect and recognize each face in the images
# Resize to QVGA so that RPI we can have acceptable fps
if out_resolution is not None:
#frame = imutils.resize(frame, width=out_resolution[0])
(h, w) = image.shape[:2]
frame = cv2.resize(
frame,
(out_resolution[0], int(h * out_resolution[0] / float(w))))
###############################################################################
# FACE DETECTION
###############################################################################
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
###############################################################################
# FACE AGE/GENDER/EMOTION ESTIMATION
###############################################################################
face_image = frame[y:y + h, h:h + w]
if model_ageestimator is not None:
age = face_age_estimator.estimate(frame, face_image)
if model_genderestimator is not None:
gender = face_gender_estimator.estimate(frame, face_image)
if model_emotionestimator is not None:
emotion = face_emotion_estimator.estimate(frame, face_image)
###############################################################################
# FACE RECOGNITION
###############################################################################
face_id, confidence = face_encoder.identify(frame, (x, y, w, h))
###############################################################################
# EYE BLINKING DETECTION
###############################################################################
total_eye_blinks, eye_counter = face_liveness.detect(
frame, (x, y, w, h), total_eye_blinks, eye_counter)
###############################################################################
# FACE POSE ESTIMATION
###############################################################################
# Detect and draw face pose locations
if model_poseestimator is not None:
shape = face_pose_estimator.detect(frame, face)
face_pose_estimator.add_overlay(frame, shape)
# Display name, age, gender, emotion
cv2.putText(frame, "Age: {}".format(age), (20, 60),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
cv2.putText(frame, "Gender: {}".format(gender), (20, 80),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
cv2.putText(frame, "Emotion: {}".format(emotion), (20, 100),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
cv2.putText(frame,
"Name: {} [{:.2f}%]".format(face_id, confidence),
(20, 120), cv2.FONT_HERSHEY_SIMPLEX, 0.5,
(255, 255, 255), 1, cv2.LINE_AA)
###############################################################################
# EYE BLINKING DETECTION
###############################################################################
cv2.putText(frame, "Blinks: {}".format(total_eye_blinks), (20, 40),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
# Update frame count
fps_frames += 1
if (framecount != 0 and fps_frames >= framecount):
break
if (fps_frames % 30 == 29):
fps = fps_frames / (time() - fps_start)
fps_frames = 0
fps_start = time()
cv2.putText(frame, "FPS {:.2f}".format(fps), (20, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
# Save the frame to a video
if saveVideo:
out.write(frame)
# Display the resulting frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
keyPressed = cv2.waitKey(1) & 0xFF
if keyPressed == 27: # ESC
break
elif keyPressed == 32: # Space
saveVideo, out = save_video(saveVideo, out, frame.shape[:2],
WINDOW_NAME + ".avi")
elif keyPressed == 13: # Enter
save_photo(
frame, WINDOW_NAME + "_" +
datetime.datetime.now().strftime("%Y%m%d_%H%M%S") + ".jpg")
# Set the fps
time_diff = time() - fps_start
if time_diff:
fps = fps_frames / time_diff
if image is not None:
cv2.waitKey(3000)
if saveVideo == True:
out.release()
# Release the camera
cam_release(cap)
return fps
def process_facerecognition():
cam_index = 0
# cam_index = "http://192.168.0.105:8000/video"
# cam_index = int(args.webcam)
cam_resolution = RESOLUTION_VGA
# model_detector=FaceDetectorModels.HAARCASCADE
# model_detector=FaceDetectorModels.DLIBHOG
# model_detector=FaceDetectorModels.DLIBCNN
# model_detector=FaceDetectorModels.SSDRESNET
model_detector = FaceDetectorModels.MTCNN
# model_detector=FaceDetectorModels.FACENET
# model_recognizer=FaceEncoderModels.LBPH
# model_recognizer=FaceEncoderModels.OPENFACE
# model_recognizer=FaceEncoderModels.DLIBRESNET
model_recognizer = FaceEncoderModels.FACENET
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
df = pd.DataFrame(columns=['Employee'])
while (True):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Identify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
# break
df = df.append({'Employee': face_id}, ignore_index=True)
df1 = df.groupby('Employee').size()
g = df1.values.tolist()
#print(g)
# Display updated frame to web app
yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' +
cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
# Release the camera
return g
camera.release()
cv2.destroyAllWindows()
def process_facerecognition(model_detector, model_recognizer,
model_speech_synthesizer, cam_index,
cam_resolution):
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize text-to-speech synthesizer
speech_synthesizer = SpeechSynthesizer(model=model_speech_synthesizer,
path=None,
path_output=None,
training=False)
except:
#face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
frame_count = 0
while (True):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Play audio file corresponding to the recognized name
if (frame_count % 30 == 0):
if len(faces) == 1 and (face_id is not None) and (face_id !=
"Unknown"):
speech_synthesizer.playaudio(INPUT_DIR_AUDIOSET,
face_id,
block=False)
# Process 1 face only
break
# Display updated frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
frame_count += 1
# Release the camera
camera.release()
cv2.destroyAllWindows()
def login():
windowname = "Result"
INPUT_DIR_DATASET = "datasets"
INPUT_DIR_MODEL_DETECTION = "models/detection/"
INPUT_DIR_MODEL_ENCODING = "models/encoding/"
INPUT_DIR_MODEL_TRAINING = "models/training/"
INPUT_DIR_MODEL_ESTIMATION = "models/estimation/"
INPUT_DIR_MODEL_LIVENESS = "models/liveness/"
# Set width and height
RESOLUTION_QVGA = (320, 240)
#cap = cv2.VideoCapture(0)
# cam_index = 0
cam_resolution = RESOLUTION_QVGA
# detector = FaceDetectorModels.HAARCASCADE
# detector=FaceDetectorModels.DLIBHOG
# detector=FaceDetectorModels.DLIBCNN
# detector=FaceDetectorModels.SSDRESNET
# detector=FaceDetectorModels.MTCNN
detector = FaceDetectorModels.FACENET
# encoder = FaceEncoderModels.LBPH
# encoder=FaceEncoderModels.OPENFACE
# encoder=FaceEncoderModels.DLIBRESNET
encoder = FaceEncoderModels.FACENET
liveness = FaceLivenessModels.EYESBLINK_MOUTHOPEN
# liveness=FaceLivenessModels.COLORSPACE_YCRCBLUV
# Initialize the camera
#camera = cam_init(cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=encoder,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize face liveness detection
face_liveness = FaceLiveness(model=liveness,
path=INPUT_DIR_MODEL_LIVENESS)
face_liveness2 = FaceLiveness(
model=FaceLivenessModels.COLORSPACE_YCRCBLUV,
path=INPUT_DIR_MODEL_LIVENESS)
except:
print("Error, check if models and trained dataset models exists!")
return
face_id, confidence = (None, 0)
eyes_close, eyes_ratio = (False, 0)
total_eye_blinks, eye_counter, eye_continuous_close = (
0, 0, 1) # eye_continuous_close should depend on frame rate
mouth_open, mouth_ratio = (False, 0)
total_mouth_opens, mouth_counter, mouth_continuous_open = (
0, 0, 1) # eye_continuous_close should depend on frame rate
time_start = time()
time_elapsed = 0
frame_count = 0
identified_unique_faces = {} # dictionary
runtime = 10 # monitor for 10 seconds only
is_fake_count_print = 0
# print("Note: this will run for {} seconds only".format(runtime))
while (True):
# Capture frame from webcam
if flask.request.method == "POST":
# image = request.get("image")
# read the image in PIL format
image = request.files["image"]
print("image :", type(image))
npimg = np.fromfile(image, np.uint8)
print("npimg :", type(npimg))
file = cv2.imdecode(npimg, cv2.IMREAD_COLOR)
print("file :", type(file))
# pil_image = Image.open(image)
#img = np.array(Image.open(io.BytesIO(image)))
# save the image on server side
# cv2.imwrite('saved_image/new.jpg', cv2.cvtColor(img, cv2.COLOR_RGB2BGR))
frame = file
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
# Check if eyes are close and if mouth is open
eyes_close, eyes_ratio = face_liveness.is_eyes_close(
frame, face)
mouth_open, mouth_ratio = face_liveness.is_mouth_open(
frame, face)
print("eyes_close={}, eyes_ratio ={:.2f}".format(
mouth_open, mouth_ratio))
print("mouth_open={}, mouth_ratio={:.2f}".format(
mouth_open, mouth_ratio))
# print("confidence: " , confidence)
# Detect if frame is a print attack or replay attack based on colorspace
is_fake_print = face_liveness2.is_fake(frame, face)
# is_fake_replay = face_liveness2.is_fake(frame, face, flag=1)
# Identify face only if it is not fake and eyes are open and mouth is close
if is_fake_print:
is_fake_count_print += 1
face_id, confidence = ("Fake", None)
elif not eyes_close and not mouth_open:
face_id, confidence = face_encoder.identify(frame, face)
if face_id not in identified_unique_faces:
identified_unique_faces[face_id] = 1
else:
identified_unique_faces[face_id] += 1
label_face(frame, face, face_id,
confidence) # Set text and bounding box on face
#cv2.imshow(windowname,frame)
#cv2.waitKey(1)
conf = confidence
id = face_id
print("confidence :", confidence)
print("faceid :", face_id)
'''if face_id in identified_unique_faces:
return render_template('success.html')
else:
return render_template('result_F.html')'''
'''POST_USERNAME = str(request.form['name'])
if POST_USERNAME==id:
return render_template('success.html')
else:
return render_template('result_F.html')'''
#yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' + cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
# Monitor eye blinking and mouth opening for liveness detection
total_eye_blinks, eye_counter = monitor_eye_blinking(
eyes_close, eyes_ratio, total_eye_blinks, eye_counter,
eye_continuous_close)
total_mouth_opens, mouth_counter = monitor_mouth_opening(
mouth_open, mouth_ratio, total_mouth_opens, mouth_counter,
mouth_continuous_open)
# Update frame count
frame_count += 1
def process_facerecognition(model_detector, model_recognizer, redis_cam, cam_resolution, embeddings_path, image_folder):
# Initialize the camera
# camera = cam_init(redis_cam, cam_resolution[0], cam_resolution[1])
if redis_cam == None:
print("cannot connect to redis streaming...")
return
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector, path=INPUT_DIR_MODEL_DETECTION)
# face_detector_fn = FaceDetector(model=FaceDetectorModels(5), path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer, path=INPUT_DIR_MODEL_ENCODING, path_training=INPUT_DIR_MODEL_TRAINING, training=False)
face_recognizer = FaceRecognizer(face_embeddings_path = embeddings_path)
except Exception as ex:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
if image_folder != '':
output_dir = './predict_results'
if not os.path.exists(output_dir):
os.makedirs(output_dir)
lst_image = glob.glob(image_folder+'/*')
for img_path in lst_image:
try:
if img_path:
image = cv2.VideoCapture(img_path)
ret, frame = image.read()
if frame is None:
print("Error, check if camera is connected!")
continue
frame, _, _ = face_recognizer.recognize(frame, face_detector, face_encoder,num_times_recognize=1, check_type = 'face_distance')
cv2.imwrite(join(output_dir, basename(img_path)), frame)
except Exception as ex:
print(ex)
else:
# input as webcam
while (True):
# Capture frame from webcam
img_path = None
# img_path = "/home/500/anh_lbt/IMAGE_TASK/FaceRecognition_2019/faceidsys/datasets/test/me2.jpg"
img_path = "/home/500/anh_lbt/IMAGE_TASK/test/test.jpg"
if img_path:
image = cv2.VideoCapture(img_path)
ret, frame = image.read()
else:
key = '{0}|uint8_{1}_{2}_3'.format((redis_cam.store_name), \
redis_cam.y1 - redis_cam.y0, redis_cam.x1 - redis_cam.x0)
frame = redis_cam.decoding_image(key)
if frame is None:
print("Error, check if camera is connected!")
break
# TEST === Detect and identify faces in the frame
# faces = face_detector.detect(frame)
# label_face(frame, faces[0], None)
# plt.imshow(frame)
# plt.show()
# embs = face_encoder.encode(frame, faces, -1)
frame, _, _ = face_recognizer.recognize(frame, face_detector, face_encoder, check_type = 'face_distance')
# Display updated frame
frame = imutils.resize(frame, 1800)
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
# Release the camera
# camera.release()
cv2.destroyAllWindows()
while (True):
# Capture frame from webcam
# ret, frame = camera.read()
key = '{0}|uint8_{1}_{2}_3'.format((redis_cam.store_name), \
redis_cam.y1 - redis_cam.y0, redis_cam.x1 - redis_cam.x0)
frame = redis_cam.decoding_image(key)
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(frame, (x, y, w, h))
# Set text and bounding box on face
label_face(frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
break
# Display updated frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
if cv2.waitKey(1) & 0xFF == 27: # ESC
break
# Release the camera
camera.release()
cv2.destroyAllWindows()
def login():
INPUT_DIR_DATASET = "datasets"
INPUT_DIR_MODEL_DETECTION = "models/detection/"
INPUT_DIR_MODEL_ENCODING = "models/encoding/"
INPUT_DIR_MODEL_TRAINING = "models/training/"
INPUT_DIR_MODEL_ESTIMATION = "models/estimation/"
INPUT_DIR_MODEL_LIVENESS = "models/liveness/"
# Set width and height
RESOLUTION_QVGA = (320, 240)
cap = cv2.VideoCapture(0)
# cam_index = 0
cam_resolution = RESOLUTION_QVGA
# detector = FaceDetectorModels.HAARCASCADE
# detector=FaceDetectorModels.DLIBHOG
# detector=FaceDetectorModels.DLIBCNN
# detector=FaceDetectorModels.SSDRESNET
# detector=FaceDetectorModels.MTCNN
detector = FaceDetectorModels.FACENET
# encoder = FaceEncoderModels.LBPH
# encoder=FaceEncoderModels.OPENFACE
# encoder=FaceEncoderModels.DLIBRESNET
encoder = FaceEncoderModels.FACENET
liveness = FaceLivenessModels.EYESBLINK_MOUTHOPEN
# liveness=FaceLivenessModels.COLORSPACE_YCRCBLUV
# Initialize the camera
#camera = cam_init(cam_resolution[0], cam_resolution[1])
try:
# Initialize face detection
face_detector = FaceDetector(model=detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=encoder,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize face liveness detection
face_liveness = FaceLiveness(model=liveness,
path=INPUT_DIR_MODEL_LIVENESS)
face_liveness2 = FaceLiveness(
model=FaceLivenessModels.COLORSPACE_YCRCBLUV,
path=INPUT_DIR_MODEL_LIVENESS)
except:
print("Error, check if models and trained dataset models exists!")
return
face_id, confidence = (None, 0)
eyes_close, eyes_ratio = (False, 0)
total_eye_blinks, eye_counter, eye_continuous_close = (
0, 0, 1) # eye_continuous_close should depend on frame rate
mouth_open, mouth_ratio = (False, 0)
total_mouth_opens, mouth_counter, mouth_continuous_open = (
0, 0, 1) # eye_continuous_close should depend on frame rate
time_start = time()
time_elapsed = 0
frame_count = 0
identified_unique_faces = {} # dictionary
runtime = 10 # monitor for 10 seconds only
is_fake_count_print = 0
# print("Note: this will run for {} seconds only".format(runtime))
while (True):
# Capture frame from webcam
ret, frame = cap.read()
if frame is None:
print("Error, check if camera is connected!")
break
# Detect and identify faces in the frame
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
# Check if eyes are close and if mouth is open
eyes_close, eyes_ratio = face_liveness.is_eyes_close(frame, face)
mouth_open, mouth_ratio = face_liveness.is_mouth_open(frame, face)
print("eyes_close={}, eyes_ratio ={:.2f}".format(
mouth_open, mouth_ratio))
print("mouth_open={}, mouth_ratio={:.2f}".format(
mouth_open, mouth_ratio))
# print("confidence: " , confidence)
# Detect if frame is a print attack or replay attack based on colorspace
is_fake_print = face_liveness2.is_fake(frame, face)
# is_fake_replay = face_liveness2.is_fake(frame, face, flag=1)
# Identify face only if it is not fake and eyes are open and mouth is close
if is_fake_print:
is_fake_count_print += 1
face_id, confidence = ("Fake", None)
elif not eyes_close and not mouth_open:
face_id, confidence = face_encoder.identify(frame, face)
if face_id not in identified_unique_faces:
identified_unique_faces[face_id] = 1
else:
identified_unique_faces[face_id] += 1
label_face(frame, face, face_id,
confidence) # Set text and bounding box on face
conf = confidence
print("confidence :", confidence)
print("faceid :", face_id)
if conf >= 90:
#return 'Recognized'
return render_template('success.html')
else:
#return 'Not recognized'
return render_template('dashboard.html')
#yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' + cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
# Monitor eye blinking and mouth opening for liveness detection
total_eye_blinks, eye_counter = monitor_eye_blinking(
eyes_close, eyes_ratio, total_eye_blinks, eye_counter,
eye_continuous_close)
total_mouth_opens, mouth_counter = monitor_mouth_opening(
mouth_open, mouth_ratio, total_mouth_opens, mouth_counter,
mouth_continuous_open)
# Update frame count
frame_count += 1
# Release the camera
cap.release()
cv2.destroyAllWindows()
def process_livenessdetection():
cam_index = 0
cam_resolution = RESOLUTION_QVGA
# model_detector=FaceDetectorModels.HAARCASCADE
model_detector = FaceDetectorModels.DLIBHOG
# model_detector=FaceDetectorModels.DLIBCNN
# model_detector=FaceDetectorModels.SSDRESNET
# model_detector=FaceDetectorModels.MTCNN
# model_detector=FaceDetectorModels.FACENET
# model_recognizer=FaceEncoderModels.LBPH
# model_recognizer=FaceEncoderModels.OPENFACE
model_recognizer = FaceEncoderModels.DLIBRESNET
# model_recognizer=FaceEncoderModels.FACENET
#liveness=FaceLivenessModels.EYESBLINK_MOUTHOPEN
liveness = FaceLivenessModels.COLORSPACE_YCRCBLUV
# Initialize the camera
camera = cam_init(cam_index, cam_resolution[0], cam_resolution[1])
check = int(
input(
"Enter number 1 if you want to show live video and vice versa: "))
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
# Initialize face liveness detection
face_liveness = FaceLiveness(
model=FaceLivenessModels.EYESBLINK_MOUTHOPEN,
path=INPUT_DIR_MODEL_LIVENESS)
face_liveness2 = FaceLiveness(
model=FaceLivenessModels.COLORSPACE_YCRCBLUV,
path=INPUT_DIR_MODEL_LIVENESS)
except:
#face_encoder = None
print("Warning, check if models and trained dataset models exists!")
return
face_id, confidence = (None, 0)
##edit
eyes_close, eyes_ratio = (False, 0)
total_eye_blinks, eye_counter, eye_continuous_close = (
0, 0, 1) # eye_continuous_close should depend on frame rate
mouth_open, mouth_ratio = (False, 0)
total_mouth_opens, mouth_counter, mouth_continuous_open = (
0, 0, 1) # eye_continuous_close should depend on frame rate
time_start = time()
time_elapsed = 0
frame_count = 0
identified_unique_faces = {} # dictionary
runtime = 1000 # monitor for 10 seconds only
is_fake_count_print = 0
is_fake_count_replay = 0
face_count = 0
##edit
time_recognition = 5
checkface = False
print("Note: this will run for {} seconds only".format(runtime))
while (time_elapsed < runtime):
# Capture frame from webcam
ret, frame = camera.read()
if frame is None:
print("Error, check if camera is connected!")
break
## Detect and identify faces in the frame
#faces = face_detector.detect(frame)
#for (index, face) in enumerate(faces):
#(x, y, w, h) = face
## Indentify face based on trained dataset (note: should run facial_recognition_training.py)
#if face_encoder is not None:
#face_id, confidence = face_encoder.identify(frame, (x, y, w, h))
## Set text and bounding box on face
#label_face(frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
#break
# Detect and identify faces in the frame
# Indentify face based on trained dataset (note: should run facial_recognition_training.py)
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
# Check if eyes are close and if mouth is open
eyes_close, eyes_ratio = face_liveness.is_eyes_close(frame, face)
mouth_open, mouth_ratio = face_liveness.is_mouth_open(frame, face)
#print("eyes_close={}, eyes_ratio ={:.2f}".format(mouth_open, mouth_ratio))
#print("mouth_open={}, mouth_ratio={:.2f}".format(mouth_open, mouth_ratio))
# Detect if frame is a print attack or replay attack based on colorspace
is_fake_print = face_liveness2.is_fake(frame, face)
is_fake_replay = face_liveness2.is_fake(frame, face, flag=1)
# Identify face only if it is not fake and eyes are open and mouth is close
if is_fake_print:
is_fake_count_print += 1
face_id, confidence = ("Fake", None)
elif is_fake_replay:
is_fake_count_replay += 1
face_id, confidence = ("Fake", None)
elif not eyes_close and not mouth_open:
face_id, confidence = face_encoder.identify(frame, face)
if (face_id
not in identified_unique_faces) & (confidence > 50):
identified_unique_faces[face_id] = 1
elif (face_id in identified_unique_faces) & (confidence > 50):
identified_unique_faces[face_id] += 1
if (face_count > 100) | (face_id == "Fake"):
face_count = 0
elif (face_id != "Fake") & (confidence > 50):
face_count += 1
print("Identifying: {:.2f} %".format((face_count / 100) * 100))
label_face(frame, face, face_id,
confidence) # Set text and bounding box on face
break # Process 1 face only
# Monitor eye blinking and mouth opening for liveness detection
total_eye_blinks, eye_counter = monitor_eye_blinking(
eyes_close, eyes_ratio, total_eye_blinks, eye_counter,
eye_continuous_close)
total_mouth_opens, mouth_counter = monitor_mouth_opening(
mouth_open, mouth_ratio, total_mouth_opens, mouth_counter,
mouth_continuous_open)
# Update frame count
frame_count += 1
time_elapsed = time() - time_start
#cv2.imshow(WINDOW_NAME, frame)
if face_count > 99:
checkface = True
break
else:
checkface = False
# Display updated frame to web app
if check == 1:
yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' +
cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
if checkface == True:
print("Hello {}!!!".format(
max(identified_unique_faces, key=identified_unique_faces.get)))
else:
print("Can not indentify your face, please try again!")
# Release the camera
camera.release()
cv2.destroyAllWindows()
def process_facerecognition(cam_resolution,
out_resolution,
framecount,
image=None,
model_detector=0,
model_recognizer=0):
# Initialize the camera
if image is not None:
cap = cv2.VideoCapture(image)
else:
cap = cam_init(cam_resolution[0], cam_resolution[1])
###############################################################################
# FACE DETECTION
###############################################################################
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION,
optimize=True)
###############################################################################
# FACE RECOGNITION
###############################################################################
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
face_id, confidence = ("Unknown", 0)
# Initialize fps counter
fps_frames = 0
fps_start = time()
fps = 0
saveVideo = False
out = None
# Optimization
skip_frames = True
skip_frames_count = 0
skip_frames_set = 2
while (True):
# Capture frame-by-frame
ret, frame = cap.read()
if ret == 0:
print("Unexpected error! " + image)
break
###############################################################################
# FACE DETECTION and FACE RECOGNITION
###############################################################################
# Detect and recognize each face in the images
# Resize to QVGA so that RPI we can have acceptable fps
if out_resolution is not None:
#frame = imutils.resize(frame, width=out_resolution[0])
(h, w) = image.shape[:2]
frame = cv2.resize(
frame,
(out_resolution[0], int(h * out_resolution[0] / float(w))))
###############################################################################
# FACE DETECTION
###############################################################################
faces = face_detector.detect(frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
###############################################################################
# FACE RECOGNITION
###############################################################################
face_id, confidence = face_encoder.identify(frame, (x, y, w, h))
# Set bounding box and text
label_face(frame, (x, y, w, h), face_id, confidence)
# Update frame count
fps_frames += 1
if (framecount != 0 and fps_frames >= framecount):
break
if (fps_frames % 30 == 29):
fps = fps_frames / (time() - fps_start)
fps_frames = 0
fps_start = time()
cv2.putText(frame, "FPS {:.2f}".format(fps), (20, 20),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1,
cv2.LINE_AA)
# Save the frame to a video
if saveVideo:
out.write(frame)
# Display the resulting frame
cv2.imshow(WINDOW_NAME, frame)
# Check for user actions
keyPressed = cv2.waitKey(1) & 0xFF
if keyPressed == 27: # ESC
break
elif keyPressed == 32: # Space
saveVideo, out = save_video(saveVideo, out, frame.shape[:2],
"facial_recognition_rpi3.avi")
# Set the fps
time_diff = time() - fps_start
if time_diff:
fps = fps_frames / time_diff
if image is not None:
cv2.waitKey(3000)
if saveVideo == True:
out.release()
# Release the camera
cam_release(cap)
return fps
def cumucount():
input_video = "tiv.mp4"
# input_video = "http://*****:*****@192.168.0.105:80/1"
# By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies.
detection_graph, category_index = backbone.set_model(
'ssd_mobilenet_v1_coco_2017_11_17')
targeted_objects = "person"
fps = 24 # change it with your input video fps
width = 640 # change it with your input video width
height = 480 # change it with your input vide height
is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects
roi = 200 # roi line position
deviation = 5 # the constant that represents the object counting area
total_passed_vehicle = 0
fourcc = cv2.VideoWriter_fourcc(*'XVID')
output_movie = cv2.VideoWriter('the_output.avi', fourcc, fps,
(width, height))
# input video
cap = cv2.VideoCapture(input_video)
cam_resolution = RESOLUTION_VGA
model_detector = FaceDetectorModels.HAARCASCADE
model_recognizer = FaceEncoderModels.LBPH
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
total_passed_vehicle = 0
counting_mode = "..."
width_heigh_taken = True
with detection_graph.as_default():
with tf.Session(graph=detection_graph) as sess:
# Definite input and output Tensors for detection_graph
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Each box represents a part of the image where a particular object was detected.
detection_boxes = detection_graph.get_tensor_by_name(
'detection_boxes:0')
# Each score represent how level of confidence for each of the objects.
# Score is shown on the result image, together with the class label.
detection_scores = detection_graph.get_tensor_by_name(
'detection_scores:0')
detection_classes = detection_graph.get_tensor_by_name(
'detection_classes:0')
num_detections = detection_graph.get_tensor_by_name(
'num_detections:0')
# for all the frames that are extracted from input video
while (cap.isOpened()):
ret, frame = cap.read()
if not ret:
print("end of the video file...")
break
input_frame = frame
# Detect and identify faces in the frame
faces = face_detector.detect(input_frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Identify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
input_frame, (x, y, w, h))
# Set text and bounding box on face
label_face(input_frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
#break
# Expand dimensions since the model expects images to have shape: [1, None, None, 3]
image_np_expanded = np.expand_dims(input_frame, axis=0)
# Actual detection
(boxes, scores, classes,
num) = sess.run([
detection_boxes, detection_scores, detection_classes,
num_detections
],
feed_dict={image_tensor: image_np_expanded})
# insert information text to video frame
font = cv2.FONT_HERSHEY_SIMPLEX
# Visualization of the results of a detection.
counter, csv_line, counting_mode = vis_util.visualize_boxes_and_labels_on_image_array_y_axis(
cap.get(1),
input_frame,
2,
is_color_recognition_enabled,
np.squeeze(boxes),
np.squeeze(classes).astype(np.int32),
np.squeeze(scores),
category_index,
targeted_objects="person",
y_reference=roi,
deviation=deviation,
use_normalized_coordinates=True,
line_thickness=4)
# when the vehicle passed over line and counted, make the color of ROI line green
if counter == 1:
cv2.line(input_frame, (0, roi), (width, roi), (0, 0xFF, 0),
5)
else:
cv2.line(input_frame, (0, roi), (width, roi), (0, 0, 0xFF),
5)
total_passed_vehicle = total_passed_vehicle + counter
# insert information text to video frame
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(
input_frame,
'Detected: ' + str(total_passed_vehicle),
(10, 35),
font,
0.8,
(0, 0xFF, 0xFF),
2,
cv2.FONT_HERSHEY_SIMPLEX,
)
cv2.putText(
input_frame,
'ROI Line',
(545, roi - 10),
font,
0.6,
(0, 0, 0xFF),
2,
cv2.LINE_AA,
)
output_movie.write(input_frame)
#print ("writing frame")
#cv2.imshow('object counting',input_frame)
#if cv2.waitKey(1) & 0xFF == ord('q'):
#break
# Display updated frame to web app
yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' +
cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
cap.release()
cv2.destroyAllWindows()
