class VideoData:
"""
Helper class to present the detected face region, landmarks and emotions.
"""
#-----------------------------------------
def __init__(self):
"""
Class constructor.
"""
self._faceDet = FaceDetector()
'''
The instance of the face detector.
'''
self._bank = GaborBank()
'''
The instance of the bank of Gabor filters.
'''
self._emotionsDet = EmotionsDetector()
'''
The instance of the emotions detector.
'''
self._face = FaceData()
'''
Data of the last face detected.
'''
self._emotions = OrderedDict()
'''
Data of the last emotions detected.
'''
#-----------------------------------------
def detect(self, frame):
"""
Detects a face and the prototypic emotions on the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to perform the detections from.
Returns
-------
ret: bool
Indication of success or failure.
"""
ret, face = self._faceDet.detect(frame)
if ret:
self._face = face
# Crop just the face region
frame, face = face.crop(frame)
# Filter it with the Gabor bank
responses = self._bank.filter(frame)
# Detect the prototypic emotions based on the filter responses
self._emotions = self._emotionsDet.detect(face, responses)
return True
else:
self._face = None
return False
#---------------------------------------------
def imprimir_tempo(self, tempo, frame, lab, val, fps):
[ano, mes, dia, hora, minuto, segundo, nano] = tempo.split('/')
nano = int(nano)
segundo = int(segundo)
minuto = int(minuto)
hora = int(hora)
dia = int(dia)
mes = int(mes)
ano = int(ano)
tempoPassado = frame * int(1000000000 / fps)
nano += tempoPassado % 1000000000
tempoPassado = int(tempoPassado / 1000000000)
if nano >= 1000000000:
nano -= 1000000000
tempoPassado += 1
segundo += tempoPassado % 60
tempoPassado = int(tempoPassado / 60)
if segundo >= 60:
segundo -= 60
tempoPassado += 1
minuto += tempoPassado % 60
tempoPassado = int(tempoPassado / 60)
if minuto >= 60:
minuto -= 60
tempoPassado += 1
hora += tempoPassado % 60
tempoPassado = int(tempoPassado / 24)
if hora >= 24:
hora -= 24
tempoPassado += 1
dia += tempoPassado % 24
if hora >= 24:
hora -= 24
saida = str(ano) + '/' + str(mes) + '/' + str(dia) + '/' + str(
hora) + '/' + str(minuto) + '/' + str(segundo) + '/' + str(
nano) + '-' + lab + '-' + val
print(saida)
#-----------------------------------------
def draw(self, frame, tempo, frameNum, fps):
"""
Draws the detected data of the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to draw the information to.
"""
empty = True
try:
face = self._face
empty = face.isEmpty()
face.draw(frame)
except:
pass
# Plot the emotion probabilities
try:
emotions = self._emotions
if empty:
labels = []
values = []
else:
labels = list(emotions.keys())
values = list(emotions.values())
bigger = labels[values.index(max(values))]
for l, v in zip(labels, values):
lab = '{}'.format(l)
val = '{:.2f}'.format(v)
self.imprimir_tempo(tempo, frameNum, lab, val, fps)
except Exception as e:
print(e)
pass
class VideoData:
"""
Helper class to present the detected face region, landmarks and emotions.
"""
#-----------------------------------------
def __init__(self):
"""
Class constructor.
"""
self._faceDet = FaceDetector()
'''
The instance of the face detector.
'''
self._bank = GaborBank()
'''
The instance of the bank of Gabor filters.
'''
self._emotionsDet = EmotionsDetector()
'''
The instance of the emotions detector.
'''
self._face = FaceData()
'''
Data of the last face detected.
'''
self._emotions = OrderedDict()
'''
Data of the last emotions detected.
'''
#-----------------------------------------
def detect(self, frame):
"""
Detects a face and the prototypic emotions on the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to perform the detections from.
Returns
-------
ret: bool
Indication of success or failure.
"""
ret, face = self._faceDet.detect(frame)
if ret:
self._face = face
# Crop just the face region
frame, face = face.crop(frame)
# Filter it with the Gabor bank
responses = self._bank.filter(frame)
# Detect the prototypic emotions based on the filter responses
self._emotions = self._emotionsDet.detect(face, responses)
return self._emotions
else:
self._face = None
return False
#-----------------------------------------
def draw(self, frame):
"""
Draws the detected data of the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to draw the information to.
"""
# Font settings
font = cv2.FONT_HERSHEY_SIMPLEX
scale = 0.5
thick = 1
glow = 3 * thick
# Color settings
black = (0, 0, 0)
white = (255, 255, 255)
yellow = (0, 255, 255)
red = (0, 0, 255)
empty = True
# Plot the face landmarks and face distance
x = 5
y = 0
w = int(frame.shape[1] * 0.2)
try:
face = self._face
empty = face.isEmpty()
#face.draw(frame)
except:
traceback.print_exc()
pass
# Plot the emotion probabilities
try:
emotions = self._emotions
if empty:
labels = []
values = []
else:
labels = list(emotions.keys())
values = list(emotions.values())
bigger = labels[values.index(max(values))]
# Draw the header
text = 'emotions'
size, _ = cv2.getTextSize(text, font, scale, thick)
y += size[1] + 20
cv2.putText(frame, text, (x, y), font, scale, black, glow)
cv2.putText(frame, text, (x, y), font, scale, yellow, thick)
y += 5
cv2.line(frame, (x, y), (x + w, y), black, 1)
size, _ = cv2.getTextSize('happiness', font, scale, thick)
t = size[0] + 20
w = 150
h = size[1]
for l, v in zip(labels, values):
lab = '{}:'.format(l)
val = '{:.2f}'.format(v)
size, _ = cv2.getTextSize(l, font, scale, thick)
# Set a red color for the emotion with bigger probability
color = red if l == bigger else yellow
y += size[1] + 15
p1 = (x + t, y - size[1] - 5)
p2 = (x + t + w, y - size[1] + h + 5)
cv2.rectangle(frame, p1, p2, black, 1)
# Draw the filled rectangle proportional to the probability
p2 = (p1[0] + int((p2[0] - p1[0]) * v), p2[1])
cv2.rectangle(frame, p1, p2, color, -1)
cv2.rectangle(frame, p1, p2, black, 1)
# Draw the emotion label
cv2.putText(frame, lab, (x, y), font, scale, black, glow)
cv2.putText(frame, lab, (x, y), font, scale, color, thick)
# Draw the value of the emotion probability
cv2.putText(frame, val, (x + t + 5, y), font, scale, black,
glow)
cv2.putText(frame, val, (x + t + 5, y), font, scale, white,
thick)
except Exception as e:
print(e)
pass
if args.fov:
fov = args.fov
f = None
elif args.f:
fov = None
f = args.f
else:
print("Please, specify FOV or focal length!")
parser.print_help()
exit(2)
step = args.step
if args.detect:
print("Will use face detector")
detector = FaceDetector()
faces = detector.find_faces(src)
masks = get_mask_from_face_detect(src, faces)
elif args.maskrcnn:
print("Will use Mask R-CNN")
from segmentation import PersonSegmentation
segm = PersonSegmentation()
segm.detect(src)
masks = segm.person_masks
elif args.mask:
print(f"Will use user-defined mask {args.mask}")
masks = get_mask_from_input(args.mask)
else:
print("Please, specify way of detection!")
parser.print_help()
exit(3)
sys.path.append('../godofeye/lib/yoloface')
from blueeyes.face_recognition import FaceDetector, FaceRecognition
from blueeyes.utils import Camera
cap = Camera(
source='rtsp://*****:*****@10.10.46.224:554/Streaming/Channels/101',
frameskip=5)
cap.start()
# Configurations
FRAME_COUNT_TO_DECIDE = 5
HOME = os.environ['HOME']
# detector = FaceDetector('mtcnn', min_face_size=60)
detector = FaceDetector('yolo', model_img_size=(128, 128))
# detector = YOLO()
recog = FaceRecognition(
# model_dir='/home/huy/code/godofeye/models',
model_dir=f'{HOME}/Downloads',
dataset=f'{HOME}/output',
vggface=False,
use_knn=True
# retrain=False
)
def process_id(result_id):
print(result_id)
class VideoData:
"""
Helper class to present the detected face region, landmarks and emotions.
"""
#-----------------------------------------
def __init__(self):
"""
Class constructor.
"""
self._faceDet = FaceDetector()
'''
The instance of the face detector.
'''
self._bank = GaborBank()
'''
The instance of the bank of Gabor filters.
'''
self._emotionsDet = EmotionsDetector()
'''
The instance of the emotions detector.
'''
self._face = FaceData()
'''
Data of the last face detected.
'''
self._emotions = OrderedDict()
'''
Data of the last emotions detected.
'''
#-----------------------------------------
def detect(self, frame):
"""
Detects a face and the prototypic emotions on the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to perform the detections from.
Returns
-------
ret: bool
Indication of success or failure.
"""
ret, face = self._faceDet.detect(frame)
if ret:
self._face = face
# Crop just the face region
frame, face = face.crop(frame)
# Filter it with the Gabor bank
responses = self._bank.filter(frame)
# Detect the prototypic emotions based on the filter responses
self._emotions = self._emotionsDet.detect(face, responses)
return True
else:
self._face = None
return False
#---------------------------------------------
def imprimir_tempo(self, tempo, frame, lab, val, fps):
[ano, mes, dia, hora, minuto, segundo, nano] = tempo.split('/')
nano = int(nano)
segundo = int(segundo)
minuto = int(minuto)
hora = int(hora)
dia = int(dia)
mes = int(mes)
ano = int(ano)
tempoPassado = frame * int(1000000000 / fps)
nano += tempoPassado % 1000000000
tempoPassado = int(tempoPassado / 1000000000)
if nano >= 1000000000:
nano -= 1000000000
tempoPassado += 1
segundo += tempoPassado % 60
tempoPassado = int(tempoPassado / 60)
if segundo >= 60:
segundo -= 60
tempoPassado += 1
minuto += tempoPassado % 60
tempoPassado = int(tempoPassado / 60)
if minuto >= 60:
minuto -= 60
tempoPassado += 1
hora += tempoPassado % 60
tempoPassado = int(tempoPassado / 24)
if hora >= 24:
hora -= 24
tempoPassado += 1
dia += tempoPassado % 24
if hora >= 24:
hora -= 24
saida = str(ano) + '/' + str(mes) + '/' + str(dia) + '/' + str(
hora) + '/' + str(minuto) + '/' + str(segundo) + '/' + str(
nano) + '-' + lab + '-' + val
print(saida)
#-----------------------------------------
def drawFrame(self, frame, labels):
atual = -1
preto = (0, 0, 0)
amarelo = (0, 255, 255)
soft = TAM_LINHA
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.line(frame, (COM_X, FIM_Y), (FIM_X, FIM_Y), preto, soft)
y = COM_Y
for l in labels:
atual += 1
lab = '{}:'.format(l)
x = OFFSETLETRA_X
y = OFFSETLETRA_FRAME - atual * TAM_LET
#size, _ = cv2.getTextSize(lab, font, 1, soft)
#maior label tem tamanho de (164,22)
#print (size)
cv2.putText(frame, lab, (x, y + OFFSETLETRA_Y), font, 1, amarelo,
soft)
cv2.line(frame, (COM_X, y), (FIM_X, y), preto, soft)
cv2.line(frame, (DIVISOR_X, y), (DIVISOR_X, FIM_Y), preto, soft)
cv2.line(frame, (FIM_X, y), (FIM_X, FIM_Y), preto, soft)
#cv2.line(frame, (600, y), (600,465), cor, soft)
return frame
#-----------------------------------------
def draw(self, frame, tempo, frameNum, vals, fps, processar):
"""
Draws the detected data of the given frame image.
Parameters
----------
frame: numpy.ndarray
Image where to draw the information to.
"""
amarelo = (0, 255, 255)
empty = True
try:
face = self._face
empty = face.isEmpty()
except:
pass
# Plot the emotion probabilities
try:
emotions = self._emotions
atual = 0
labels = [
'Neutral', 'Felicidade', 'Tristeza', 'Raiva', 'Medo',
'Surpresa', 'Desgosto'
]
if empty:
values = [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
else:
values = list(emotions.values())
bigger = labels[values.index(max(values))]
frame = self.drawFrame(frame, labels)
for l, v in zip(labels, values):
lab = '{}'.format(l)
val = '{:.2f}'.format(v)
vals[atual].rotate(-1)
vals[atual].pop()
vals[atual].append(v)
for i in range(PONTOS - 1):
valor1 = int(OFFSETPONTO - vals[atual][i] * RESOL_LINHA_Y -
atual * RESOL_LINHA_Y)
valor2 = int(OFFSETPONTO -
vals[atual][i + 1] * RESOL_LINHA_Y -
atual * RESOL_LINHA_Y)
cv2.line(frame, (OFFSETLINHA + RESOL_LINHA_X * i, valor1),
(OFFSETLINHA + RESOL_LINHA_X * (i + 1), valor2),
amarelo, TAM_LINHA)
#cv2.putText(frame, val, (5, 20 + atual*25), font, 1, yellow, 1)
#cv2.putText(frame, '{}'.format(vals[atual][199]), (320, 20 + atual*25), font, 1, yellow, 1)
if processar:
self.imprimir_tempo(tempo, frameNum, lab, val, fps)
atual += 1
return frame, vals
except Exception as e:
print(e)
pass