def ff(img):
"""
:param img:
:return: faceAligned[0] the aligned face version of the original input image
"""
# initialize dlib's face detector (HOG-based) and then create
# the facial landmark predictor and the face aligner
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(str(_path) + "/shape_predictor_68_face_landmarks.dat") ########################
fa = FaceAligner(predictor, desiredFaceWidth=256)
# load the input image, resize it, and convert it to grayscale
image = img ##########################################################
image = imutils.resize(image, width=800)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# show the original input image and detect faces in the grayscale
# image
#cv2.imshow("Input", image)
rects = detector(gray, 2)
lst = []
# loop over the face detections
for rect in rects:
# extract the ROI of the *original* face, then align the face
# using facial landmarks
(x, y, w, h) = rect_to_bb(rect)
faceOrig = imutils.resize(image[y:y + h, x:x + w], width=256)
faceAligned = fa.align(image, gray, rect)
import uuid
f = str(uuid.uuid4())
cv2.imwrite("foo/" + f + ".png", faceAligned)
# display the output images
#cv2.imshow("Original", faceOrig)
#cv2.imshow("Aligned", faceAligned)
#cv2.waitKey(0)
lst.append(faceAligned)
return lst[0]
if not is_exist_user:
user_data.append(entry)
user_data = sorted(user_data, key=lambda user_key: user_key['id'], reverse=False)
print("[DEBUG] user_date with index = {}".format(user_data))
with open(user_data_file, mode='w') as output_file:
output_file.write(json.dumps(user_data, indent=4))
output_file.close()
break
# if the `s` key was pressed save the first found face
if key == ord('s'):
if len(faces) > 0:
face_aligned = fa.align(frame, gray_frame, faces[0])
image_name = "{}{}.png".format(base_dir, str(datetime.datetime.now()).replace(" ", "_").replace(":", "."))
print("[DEBUG] image_name = {}".format(image_name))
# save image
cv2.imwrite(image_name, face_aligned)
# show image
cv2.imshow(image_name, face_aligned)
# loop over the face detections
for face in faces:
(x, y, w, h) = face_utils.rect_to_bb(face)
cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
# show the frame
cv2.imshow("Face Detection Window - S: Save/Capture, Q: Quit", frame)
def test():
# 取得dlib預設的臉部偵測器
detector = dlib.get_frontal_face_detector()
# 根據shape_predictor方法載入68個特徵點模型，此方法為人臉表情識別的偵測器
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
# 臉部校正
fa = FaceAligner(predictor, desiredFaceWidth=256)
# 載入人臉辨識檢測器
facerec = dlib.face_recognition_model_v1(
"dlib_face_recognition_resnet_model_v1.dat")
# 比對人臉描述子列表
descriptors = []
# 比對人臉名稱列表
candidate = []
# 開啟影片檔案
cap = cv2.VideoCapture(0, cv2.CAP_DSHOW)
# 比對人臉圖片資料夾名稱
faces_folder_path = "./rec"
WHO = None
Erro = 0
# 有讀過的照片存成新的檔可直接使用，有新的圖片再加
# 有讀過的照片存成新的檔可直接使用，有新的圖片再加
# 路徑中的file個數
NUM_OF_FILES = 0 # 檔案數
for fn in os.listdir(faces_folder_path):
NUM_OF_FILES += 1
# print(NUM_OF_FILES)
person = [0 for i in range(NUM_OF_FILES)]
# print(person)
# 原本的個數
f = open("numOfFiles.txt", encoding="utf-8")
num_o = f.read()
# print(type(num_o))
num_o = int(num_o)
# print(type(num_o))
# print(num_o)
if num_o != NUM_OF_FILES:
# 把新的個數寫到txt檔裡
num = open('numOfFiles.txt', 'w', encoding="utf-8")
num.write(str(NUM_OF_FILES))
num.close()
# 跑新的
# 讀取資料夾裡的圖片及檔案名(人名)，並將每張圖的128維特徵向量存到description一維陣列中
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
base = os.path.basename(f)
# 依序取得圖片檔案人名，存到candidate一維陣列中
candidate.append(os.path.splitext(base)[0])
img = io.imread(f)
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
# 1.人臉偵測
face_rects = detector(img, 0)
# print(face_rects)
for index, face in enumerate(face_rects):
# 2.人臉校正
faceAligned = fa.align(img, gray, face)
# 3.再人臉偵測(去除校正後多餘的部分)
face_rects2 = detector(faceAligned, 1)
for index2, face2 in enumerate(face_rects2):
ax1 = face2.left()
ay1 = face2.top()
ax2 = face2.right()
ay2 = face2.bottom()
# 4.68特徵點偵測
shape = predictor(faceAligned, face2)
# 5.取得描述子，128維特徵向量
face_descriptor = facerec.compute_face_descriptor(
faceAligned, shape)
# 轉換numpy array格式
v = np.array(face_descriptor)
descriptors.append(v)
# print(face_rects)
# 存新的到原本的txt檔
np.savetxt('desc_txt.txt', descriptors)
c = open("cand.txt", "w", encoding="utf-8")
str_list = [line + '\n' for line in candidate]
c.writelines(str_list)
c.close()
# 以迴圈從影片檔案讀取影格，並顯示出來
while cap.isOpened():
desc_list = np.loadtxt("./desc_txt.txt")
k = open("cand.txt", encoding="utf-8")
can = k.readlines()
x1 = 0
x2 = 0
y1 = 0
y2 = 0
# 從視訊鏡頭擷取畫面
ret, frame = cap.read()
# 縮小圖片
frame = imutils.resize(frame, width=800)
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
# 1.人臉偵測
face_rects = detector(frame, 1)
# 取出所有偵測的結果(所有人臉座標點)
for index, rect in enumerate(face_rects):
x1 = rect.left()
y1 = rect.top()
x2 = rect.right()
y2 = rect.bottom()
# 以方框標示偵測的人臉
cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 4,
cv2.LINE_AA)
# 2.人臉校正
faceAligned = fa.align(frame, gray, rect)
# 3.再人臉偵測(去除校正後多餘的部分)
face_rects2 = detector(faceAligned, 1)
# 取出所有偵測的結果(所有人臉座標點)
for index2, rect2 in enumerate(face_rects2):
ax1 = rect2.left()
ay1 = rect2.top()
ax2 = rect2.right()
ay2 = rect2.bottom()
# 4.68特徵點偵測
shape = predictor(faceAligned, rect2)
# 5.取得描述子，128維特徵向量
face_descriptor = facerec.compute_face_descriptor(
faceAligned, shape)
# 轉換numpy array格式
d_test = np.array(face_descriptor)
# 計算歐式距離 (與圖片庫裡，各個人臉間的距離)(5張照片就有5個距離)
# 清空 存放人臉距離的陣列
dist = []
for index in desc_list:
# 計算距離
dist_ = np.linalg.norm(index - d_test)
# 加入陣列
dist.append(dist_)
# print(dist)
# 辨識人名
if dist != []:
# 將比對人名和比對出來的歐式距離組成一個dict
c_d = dict(zip(can, dist))
# 根據歐式距離由小到大排序 [("名字",距離)]二微陣列
cd_sorted = sorted(c_d.items(), key=lambda d: d[1])
# print(cd_sorted)
# 歐式距離(0~1)越小越像，設定0.5作為最低辨識標準
if cd_sorted[0][1] < 0.4:
rec_name = cd_sorted[0][0]
else:
rec_name = "No Data"
Erro = Erro + 1
if Erro == 6:
WHO = "No Data"
cap.release()
cv2.destroyAllWindows()
break
imgPil = Image.fromarray(frame)
font = ImageFont.truetype("C:/Windows/Fonts/msjh.ttc", 20)
draw = ImageDraw.Draw(imgPil)
draw.fontmode = '1' # 關閉反鋸齒
draw.text((x1, y1 - 20),
rec_name,
font=font,
fill=(255, 255, 255))
frame = np.array(imgPil)
for index, name in enumerate(can):
if rec_name == name:
person[index] = person[index] + 1
# print(person)
if person[index] == 4:
WHO = can[index]
# 標示辨識的人名(中文)
if WHO is not None:
# imgPil.show()
print(WHO)
time.sleep(5)
cap.release()
cv2.destroyAllWindows()
return WHO
break
elif sum(person) > 8:
WHO = "No Data"
cap.release()
cv2.destroyAllWindows()
break
# 標示辨識的人名(只能標示英文)
# cv2.putText(frame, rec_name, (x1, y1), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2, cv2.LINE_AA)
# CV2是用BGR
frame = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
"""
# FPS
# Find OpenCV version
(major_ver, minor_ver, subminor_ver) = (cv2.__version__).split('.')
if int(major_ver) < 3:
fps = cap.get(cv2.cv.CV_CAP_PROP_FPS)
print("Frames per second using video.get(cv2.cv.CV_CAP_PROP_FPS): {0}".format(fps))
else:
fps = cap.get(cv2.CAP_PROP_FPS)
print("Frames per second using video.get(cv2.CAP_PROP_FPS) : {0}".format(fps))
"""
# 顯示結果
cv2.imshow("Face Detection", frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
count = 0
while number_of_images < MAX_NUMBER_OF_IMAGES:
ret, frame = video_capture.read()
frame = cv2.flip(frame, 1)
frame_gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
#faces = face_cascade.detectMultiScale(frame, 1.3, 5)
faces = detector(frame_gray)
if len(faces) == 1:
face = faces[0]
(x, y, w, h) = face_utils.rect_to_bb(face)
face_img = frame_gray[y - 50:y + h + 100, x - 50:x + w + 100]
face_aligned = face_aligner.align(frame, frame_gray, face)
if count == 5:
cv2.imwrite(
os.path.join(directory,
str(name + "_" + str(number_of_images) + '.jpg')),
face_aligned)
number_of_images += 1
count = 0
print(count)
count += 1
cv2.imshow('Video', frame)
if (cv2.waitKey(1) & 0xFF == ord('q')):
break
right = bb[0] + bb[2]
bottom = bb[1] + bb[3]
return np.array([top, right, bottom, left])
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor('shape_predictor_68_face_landmarks.dat')
fa = FaceAligner(predictor, desiredFaceWidth=256)
cap = cv2.VideoCapture(0)
while (True):
ret, frame = cap.read()
cv2.imshow('frame', frame)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
rects = detector(gray, 2)
for rect in rects:
(x, y, w, h) = rect_to_bb(rect)
faceOrig = imutils.resize(frame[y:y + h, x:x + w], width=256)
faceAligned = fa.align(frame, gray, rect)
faceAligned = fa.align(frame, gray, rect)
cv2.imshow("Original", faceOrig)
cv2.imshow("Aligned", faceAligned)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def create_dataset(username):
id = username
if (os.path.exists(
'face_recognition_data/training_dataset/{}/'.format(id)) == False):
os.makedirs('face_recognition_data/training_dataset/{}/'.format(id))
directory = 'face_recognition_data/training_dataset/{}/'.format(id)
# Detect face
#Loading the HOG face detector and the shape predictpr for allignment
print("[INFO] Loading the facial detector")
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(
'face_recognition_data/shape_predictor_68_face_landmarks.dat'
) #Add path to the shape predictor ######CHANGE TO RELATIVE PATH LATER
fa = FaceAligner(predictor, desiredFaceWidth=96)
#capture images from the webcam and process and detect the face
# Initialize the video stream
print("[INFO] Initializing Video stream")
vs = VideoStream(src=0).start()
#time.sleep(2.0) ####CHECK######
# Our identifier
# We will put the id here and we will store the id with a face, so that later we can identify whose face it is
# Our dataset naming counter
sampleNum = 0
# Capturing the faces one by one and detect the faces and showing it on the window
while (True):
# Capturing the image
#vs.read each frame
frame = vs.read()
#Resize each image
frame = imutils.resize(frame, width=800)
#the returned img is a colored image but for the classifier to work we need a greyscale image
#to convert
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
#To store the faces
#This will detect all the images in the current frame, and it will return the coordinates of the faces
#Takes in image and some other parameter for accurate result
faces = detector(gray_frame, 0)
#In above 'faces' variable there can be multiple faces so we have to get each and every face and draw a rectangle around it.
for face in faces:
print("inside for loop")
(x, y, w, h) = face_utils.rect_to_bb(face)
face_aligned = fa.align(frame, gray_frame, face)
# Whenever the program captures the face, we will write that is a folder
# Before capturing the face, we need to tell the script whose face it is
# For that we will need an identifier, here we call it id
# So now we captured a face, we need to write it in a file
sampleNum = sampleNum + 1
# Saving the image dataset, but only the face part, cropping the rest
if face is None:
print("face is none")
continue
cv2.imwrite(directory + '/' + str(sampleNum) + '.jpg',
face_aligned)
face_aligned = imutils.resize(face_aligned, width=400)
#cv2.imshow("Image Captured",face_aligned)
# @params the initial point of the rectangle will be x,y and
# @params end point will be x+width and y+height
# @params along with color of the rectangle
# @params thickness of the rectangle
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 1)
# Before continuing to the next loop, I want to give it a little pause
# waitKey of 100 millisecond
cv2.waitKey(50)
#Showing the image in another window
#Creates a window with window name "Face" and with the image img
cv2.imshow("Add Images", frame)
#Before closing it we need to give a wait command, otherwise the open cv wont work
# @params with the millisecond of delay 1
cv2.waitKey(1)
#To get out of the loop
if (sampleNum > 300):
break
#Stoping the videostream
vs.stop()
# destroying all the windows
cv2.destroyAllWindows()
def register_yourself(student_id, frame_num, image_num, id_idx):
try:
os.makedirs(DATA_PATH)
except:
pass
#Loading the stored face encodings and corresponding IDs
try:
with open(os.path.join(STORAGE_PATH, "known_face_ids.pickle"),
"rb") as fp:
known_face_ids = pickle.load(fp)
with open(os.path.join(STORAGE_PATH, "known_face_encodings.pickle"),
"rb") as fp:
known_face_encodings = pickle.load(fp)
except:
known_face_encodings = []
known_face_ids = []
mpl.rcParams['toolbar'] = 'None'
print("[INFO] Loading Face Detector")
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(LANDMARK_PATH)
fa = FaceAligner(predictor, desiredFaceWidth=96)
print("[INFO] Initializing Video stream")
vs = cv2.VideoCapture(0) #,cv2.CAP_DSHOW)
#Uncommnet below to create the folder for the images
'''
IMAGE_PATH = os.path.join(DATA_PATH, student_id)
try:
os.makedirs(IMAGE_PATH)
except:
pass
'''
#Entry time
tin = time.time()
#frame = vs.read()
#fig = plt.figure()
#plot = plt.subplot(1,1,1)
#plt.title("Detecting Face")
#plt.axis('off')
#im1 = plot.imshow(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
while image_num < 10: # Take 10 images
_, frame = vs.read()
frame_num += 1
#Resize each image
#frame = cv2.resize(frame ,(600,600))
#Applying face enhancement steps
frame = imageEnhancement.adjust_gamma(frame, gamma=1.5)
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
#Detecting faces in the frame
faces = detector(gray_frame, 0)
for face in faces:
if face is None:
print("face is none")
continue
#Capture the face and align it using the face aligner
(x, y, w, h) = rect_to_bb(face)
face_aligned = fa.align(frame, gray_frame, face)
face_aligned = cv2.resize(face_aligned, (600, 600))
# @params the initial point of the rectangle will be x,y and
# @params end point will be x+width and y+height
# @params along with color of the rectangle
# @params thickness of the rectangle
#Put a bounding box over detected face
frame = cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0),
1)
#cv2.imshow("Image Captured",frame)
#cv2.waitKey(50)
#plt.ion()
#im1.set_data(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
#plt.pause(0.001)
#plt.show()
if (frame_num % 30 == 0):
#Uncommnet the line below to store the face images
#cv2.imwrite(IMAGE_PATH + "/{}_".format(student_id) + str(j) + ".jpg", face_aligned)
#Appending the face encodings and corresponding IDs
try:
known_face_encodings.append(
face_recognition.face_encodings(frame)[0])
known_face_ids.append(student_id)
except:
continue
image_num += 1
#OpenCV's implementation to show an image in window(doesn't work on production server)
#cv2.imshow("Capturing Images for registration (PRESS Q TO QUIT",frame)
#Encoding the frame to be stream into browser
frame = cv2.imencode('.jpg', frame)[1].tobytes()
yield (b'--frame\r\n'
b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')
#if(cv2.waitKey(1) == ord("q")):
# break
#Storing the face encodings and corresponding IDs to disk
with open(os.path.join(STORAGE_PATH, "known_face_ids.pickle"), "wb") as fp:
pickle.dump(known_face_ids, fp)
with open(os.path.join(STORAGE_PATH, "known_face_encodings.pickle"),
"wb") as fp:
pickle.dump(known_face_encodings, fp)
#Noting the number of pictures already captured and storing the index
id_idx[student_id] = image_num
with open(os.path.join(STORAGE_PATH, "id_idx.json"), "w") as outfile:
json.dump(id_idx, outfile)
#Exit time
tout = time.time()
print(tout - tin)
#plt.close()
#Releasing the videostream
vs.release()
cv2.destroyAllWindows()
return True
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor("facerec.dat")
fa = FaceAligner(predictor, desiredFaceWidth=256)
#Load Input
al_image = cv2.imread(args["image"])
al_image = imutils.resize(al_image, width=800)
al_gray = cv2.cvtColor(al_image, cv2.COLOR_BGR2GRAY)
al_rects = detector(al_gray, 2)
# loop over the face detections
for rect in al_rects:
#Format Conversion DLIB and openCV, store new image in faceAligned
(x, y, w, h) = rect_to_bb(rect)
faceAligned = fa.align(al_image, al_gray, rect)
#now that we have the fixed image, lets get our facial landmarks
image = faceAligned
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
rects = detector(gray, 1)
#Getting the right region for the face
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = face_utils.shape_to_np(shape)
#Looping over the coordinates for the facial landmarks and display them on image
for (x, y) in shape:
cv2.circle(image, (x, y), 1, (0, 0, 255), -1)
M[1, 2] += (tY - eyesCenter[1])
# apply the affine transformation
(w, h) = (self.desiredFaceWidth, self.desiredFaceHeight)
output = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_CUBIC)
# apply the affine transformation
(w, h) = (self.desiredFaceWidth, self.desiredFaceHeight)
output = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_CUBIC)
print(img_name + " was aligned.")
cv2.imwrite('../data/aligned/' + 'aligned_' + str(c) + img_name,
output)
return
fa = FaceAligner(desiredFaceWidth=256)
def wrap(row):
fa.align(row['urls'], row['face_coords'])
if len(sys.argv) == 1:
path = '../data/images/'
files = os.listdir(path)
for img_name in files:
ans = fa.align(img_name)
cv2.imwrite('../data/aligned/' + 'aligned_' + str(c) + img_name,
output)
else:
data = pd.read_csv(sys.argv[1], encoding='utf-8')
clusters = data['cluster']
imgs = data.apply(wrap, axis=1)
# compute the bounding box of the face and draw it on the
# frame
(bX, bY, bW, bH) = face_utils.rect_to_bb(rect)
cv2.rectangle(frame, (bX, bY), (bX + bW, bY + bH), (0, 255, 0), 1)
#cropped_image = frame[bX:bX+5*bH,bY:bY+5*bW]
#cv2.namedWindow("cropped image",2)
#cv2.imshow("cropped image",cropped_image)
# determine the facial landmarks for the face region, then
# convert the facial landmark (x, y)-coordinates to a NumPy
# array
shape = predictor(gray_frame, rect)
shape = face_utils.shape_to_np(shape)
if cv2.waitKey(1) & 0xFF == ord('c'):
faceAligned = fa.align(dup_frame, gray_frame, rect)
capture = "image_" + str(randint(0, 125)) + "_" + str(
randint(0, 125))
cv2.putText(frame, str("Capturing Image"), (x - 10, y - 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 255, 255), 4)
cv2.imwrite(
"C:/Users/rai_a/PycharmProjects/final_year_face_recogn/aligned_image_data/"
+ capture + ".jpg", faceAligned)
print("Capturing in process", end="", flush=True)
time.sleep(2)
print("\nCapture Complete")
cv2.putText(frame, str("Capture Complete, Starting Processing"),
(x - 10, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.8,
(255, 255, 255), 4)
cv2.destroyAllWindows()
def main():
global Result, waitting, current_face
threads = []
waitting = False # can be passed to the TF
period = 0
args = get_args()
depth = args.depth
k = args.width
# for face detection
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
fa = FaceAligner(predictor, desiredFaceWidth=200)
# load model and weights
img_size = 200
# capture video
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 640)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)
while True:
# get video frame
ret, img = cap.read()
if not ret:
print("error: failed to capture image")
return -1
input_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
input_img2 = input_img.astype(np.float32)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
img_h, img_w, _ = np.shape(input_img)
# detect faces using dlib detector
detected = detector(input_img, 1)
faces = np.empty((len(detected), img_size, img_size, 3))
for i, d in enumerate(detected):
x1, y1, x2, y2, w, h = d.left(), d.top(
), d.right() + 1, d.bottom() + 1, d.width(), d.height()
xw1 = max(int(x1 - 0.4 * w), 0)
yw1 = max(int(y1 - 0.4 * h), 0)
xw2 = min(int(x2 + 0.4 * w), img_w - 1)
yw2 = min(int(y2 + 0.4 * h), img_h - 1)
cv2.rectangle(img, (x1, y1), (x2, y2), (255, 0, 0), 2)
# cv2.rectangle(img, (xw1, yw1), (xw2, yw2), (255, 0, 0), 2)
faces[i, :, :, :] = fa.align(input_img, gray, detected[i])
# faces[i,:,:,:] = cv2.resize(img[yw1:yw2 + 1, xw1:xw2 + 1, :], (img_size, img_size))
# add = 'tmp_{}.jpg'.format(i)
#
#cv2.imwrite('tmp_{}.jpg'.format(i), faces, [int(cv2.IMWRITE_JPEG_QUALITY), 90])
#cv2.imwrite('tmp_{}.jpg'.format(i), faces, [int(cv2.IMWRITE_JPEG_QUALITY), 90])
if len(detected) > 0:
period += 1
if not waitting:
race = np.zeros(len(detected))
gender = np.zeros(len(detected))
if not waitting and period == 10: # first time a person come to camera
period = 0
waitting = True
# image = align_face(img)
# predict ages and genders of the detected faces
for i, d in enumerate(detected):
# gender[i], race[i] = predictor(faces[i, :, :, :])
cv2.imwrite('tmp_{}.jpg'.format(i), faces[i, :, :, :],
[int(cv2.IMWRITE_JPEG_QUALITY), 180])
# cv2.imwrite('tmp_{}.jpg'.format(i), image, [int(cv2.IMWRITE_JPEG_QUALITY), 90])
# cv2.imwrite('tmp_{}.jpg'.format(i), gray, [int(cv2.IMWRITE_JPEG_QUALITY), 90])
# gender[i], race[i] = predictor2(img)
imgp = './tmp_{}.jpg'.format(i)
image_tru = load_image(imgp)
# predictor2(image_tru, len(detected))
thread = threading.Thread(target=predictor2,
args=(
image_tru,
len(detected),
))
# if thread.isAlive():
threads.append(thread)
print('thread created!')
current_face = 0
thread.start()
if Result:
gender[current_face] = Result[0] + 1
race[current_face] = Result[1] + 1
print(race, gender)
current_face += 1
Result = None
if len(detected) == 0:
waitting = False
Result = None
period = 0
threads = []
# draw results
PRase = ['-', 'W', 'B', 'A', 'I', 'O']
PGender = ['-', 'M', 'F']
for i, d in enumerate(detected):
label = "{}, {}".format(PRase[int(race[i])],
PGender[int(gender[i])])
draw_label(img, (d.left(), d.top()), label)
cv2.imshow("result", img)
key = cv2.waitKey(1)
if key == 27:
break
def main(model):
"""
"""
depth = 16
k = 8
# for face detection
detector = dlib.get_frontal_face_detector()
# tt = time.time()
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
fa = FaceAligner(predictor, desiredFaceWidth=160)
# print("time align: ", time.time() - tt)
# print(fa)
# print(type(fa))
# load model and weights
img_size = 160
# capture video
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1140)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 980)
start_time = time.time()
count_frame = 0
while True:
# get video frame
t1 = time.time()
ret, img = cap.read()
if not ret:
print("error: failed to capture image")
return -1
input_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
img_h, img_w, _ = np.shape(input_img)
# detect faces using dlib detector
detected = detector(input_img, 1)
# print(detected)
# print(type(detected))
faces = np.empty((len(detected), img_size, img_size, 3))
for i, d in enumerate(detected):
x1, y1, x2, y2, w, h = d.left(), d.top(), d.right() + 1, d.bottom() + 1, d.width(), d.height()
xw1 = max(int(x1 - 0.4 * w), 0)
yw1 = max(int(y1 - 0.4 * h), 0)
xw2 = min(int(x2 + 0.4 * w), img_w - 1)
yw2 = min(int(y2 + 0.4 * h), img_h - 1)
cv2.rectangle(img, (x1, y1), (x2, y2), (255, 0, 0), 2)
cv2.rectangle(img, (xw1, yw1), (xw2, yw2), (255, 0, 0), 2)
# print(img.shape)
# cv2.imwrite("output/test_{}.jpg".format(i), img)
faces[i, :, :, :] = fa.align(input_img, gray, detected[i])
# faces[i,:,:,:] = cv2.resize(img[yw1:yw2 + 1, xw1:xw2 + 1, :], (img_size, img_size))
# estimate age, gender
if len(detected) > 0:
gender_list = []
age_list = []
print("shape:", faces.shape)
print("len: ", len(faces))
faces = faces.astype('float32')/255
gender_arr, age_arr = model.predict(faces)
# predict gender
for gen in gender_arr:
gender_list.append(np.argmax(gen))
# predict age
for age in age_arr:
age_list.append(np.argmax(age))
# save_age.append(age_list)
# save_gender.append(gender_list)
t2 = time.time()
print("gender: ",gender_list)
print("age: ", age_list)
print("time: ", t2-t1)
time_ = (time.time()-start_time)
if time_ > sec:
path_out = "output"+"/{}.csv".format(date.today())
if not os.path.exists(path_out):
gen_csv(path_out)
# save_csv(gender_list, age_list, path_out)
# print("dump....")
start_time = time.time()
count_frame_id = 0
# draw results
for i, d in enumerate(detected):
__age = str(age_list[i])
draw_age = label_age[__age]
label = "{}, {}, ID={}".format(draw_age, "F" if gender_list[i] == 0 else "M", i)
draw_label(img, (d.left(), d.top()), label)
cv2.imshow("Output of predict", img)
key = cv2.waitKey(1)
if key == 27:
break
def cameraCap(self):
self.uId = self.userId.get()
if self.uId != '':
if self.outputLabel != None:
self.outputLabel.destroy()
self.outputLabel = Label(self.framePic, text="Here We Start")
self.outputLabel.config(font=("Courier", 44))
self.outputLabel.place(x=400, y=100)
mark_detector = MarkDetector()
name = self.uId
directory = os.path.join(facepath, name)
if not os.path.exists(facepath):
os.makedirs(facepath, exist_ok='True')
if not os.path.exists(directory):
try:
os.makedirs(directory, exist_ok='True')
except OSError as e:
if e.errno != errno.EEXIST:
print('invalid student id or access denied')
return
poses = ['frontal', 'right', 'left', 'up', 'down']
file = 0
ret, sample_frame = self.cap.read()
i = 0
count = 0
if ret == False:
return
# Introduce pose estimator to solve pose. Get one frame to setup the
# estimator according to the image size.
height, width = sample_frame.shape[:2]
pose_estimator = PoseEstimator(img_size=(height, width))
# Introduce scalar stabilizers for pose.
pose_stabilizers = [
Stabilizer(state_num=2,
measure_num=1,
cov_process=0.1,
cov_measure=0.1) for _ in range(6)
]
images_saved_per_pose = 0
number_of_images = 0
shape_predictor = dlib.shape_predictor(
"shape_predictor_68_face_landmarks.dat")
face_aligner = FaceAligner(shape_predictor,
desiredFaceWidth=FACE_WIDTH)
while i < 5:
saveit = False
# Read frame, crop it, flip it, suits your needs.
ret, frame = self.cap.read()
if ret is False:
break
if count % 5 != 0: # skip 5 frames
count += 1
continue
if images_saved_per_pose == IMAGE_PER_POSE:
i += 1
images_saved_per_pose = 0
# If frame comes from webcam, flip it so it looks like a mirror.
if file == 0:
frame = cv2.flip(frame, 2)
original_frame = frame.copy()
frame_gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
facebox = mark_detector.extract_cnn_facebox(frame)
if facebox is not None:
# Detect landmarks from image of 128x128.
x1 = max(facebox[0] - 0, 0)
x2 = min(facebox[2] + 0, width)
y1 = max(facebox[1] - 0, 0)
y2 = min(facebox[3] + 0, height)
face = frame[y1:y2, x1:x2]
face_img = cv2.resize(face,
(CNN_INPUT_SIZE, CNN_INPUT_SIZE))
face_img = cv2.cvtColor(face_img, cv2.COLOR_BGR2RGB)
marks = mark_detector.detect_marks([face_img])
# Convert the marks locations from local CNN to global image.
marks *= (facebox[2] - facebox[0])
marks[:, 0] += facebox[0]
marks[:, 1] += facebox[1]
# Try pose estimation with 68 points.
pose = pose_estimator.solve_pose_by_68_points(marks)
# Stabilize the pose.
steady_pose = []
pose_np = np.array(pose).flatten()
for value, ps_stb in zip(pose_np, pose_stabilizers):
ps_stb.update([value])
steady_pose.append(ps_stb.state[0])
steady_pose = np.reshape(steady_pose, (-1, 3))
# print(steady_pose[0][0])
# if steady_pose[0][0]>0.1:
# print('right')
# else:
# if steady_pose[0][0]<-0.1:
# print('left')
# if steady_pose[0][1]>0.1:
# print('down')
# else:
# if steady_pose[0][1]<-0.1:
# print('up')
# print(steady_pose[0])
if i == 0:
if abs(steady_pose[0][0]) < ANGLE_THRESHOLD and abs(
steady_pose[0][1]) < ANGLE_THRESHOLD:
images_saved_per_pose += 1
saveit = True
if i == 1:
if steady_pose[0][0] > ANGLE_THRESHOLD:
images_saved_per_pose += 1
saveit = True
if i == 2:
if steady_pose[0][0] < -ANGLE_THRESHOLD:
images_saved_per_pose += 1
saveit = True
if i == 3:
if steady_pose[0][1] < -ANGLE_THRESHOLD:
images_saved_per_pose += 1
saveit = True
if i == 4:
if steady_pose[0][1] > ANGLE_THRESHOLD:
images_saved_per_pose += 1
saveit = True
# Show preview.
if i >= 5:
print('Thank you')
if self.outputLabel != None:
self.outputLabel.destroy()
self.outputLabel = Label(self.framePic, text="Thanks")
self.outputLabel.config(font=("Courier", 44))
self.outputLabel.place(x=400, y=100)
break
frame = cv2.putText(
frame, poses[i] + ' : ' + str(images_saved_per_pose) +
'/' + str(IMAGE_PER_POSE), (10, 50),
cv2.FONT_HERSHEY_SIMPLEX, 2, (255, 255, 255), 1,
cv2.LINE_AA)
frame = cv2.rectangle(frame, (x1, y1), (x2, y2),
(255, 255, 0), 2)
frame = imutils.resize(frame, width=300, height=300)
# OpenCV represents images in BGR order; however PIL
# represents images in RGB order, so we need to swap
# the channels, then convert to PIL and ImageTk format
image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
image = Image.fromarray(image)
image = ImageTk.PhotoImage(image)
# if the panel is not None, we need to initialize it
if self.panel is None:
self.panel = Label(self.framePic, image=image)
self.panel.image = image
self.panel.pack(side=LEFT, padx=0, pady=0)
print("Done")
# otherwise, simply update the panel
else:
self.panel.configure(image=image)
self.panel.image = image
if saveit:
face = dlib.rectangle(x1, y1, x2, y2)
face_aligned = face_aligner.align(
original_frame, frame_gray, face)
cv2.imwrite(
os.path.join(
directory,
str(name) + '_' + str(number_of_images) +
'.jpg'), face_aligned)
print(images_saved_per_pose)
number_of_images += 1
self.cap.release()
else:
if self.outputLabel != None:
self.outputLabel.destroy()
self.outputLabel = Label(self.framePic,
text="Please Enter a Valid Id")
self.outputLabel.config(font=("Courier", 44))
self.outputLabel.place(x=400, y=100)
def upload_file():
file = request.files['image']
image_path = os.path.sep.join([UPLOAD_FOLDER, file.filename])
file.save(image_path)
# image_url = uploader.upload(image_path)
# image = AgeGenderHelper.url_to_image(image_url['url'])
# initialize dlib's face detector (HOG-based), then create facial landmark predictor and face aligner
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(deploy.DLIB_LANDMARK_PATH)
fa = FaceAligner(predictor)
# initialize image preprocessors
sp, cp, iap = SimplePreprocessor(
256, 256, inter=cv2.INTER_CUBIC), CropPreprocessor(
config.IMAGE_SIZE, config.IMAGE_SIZE,
horiz=False), ImageToArrayPreprocessor()
# loop over image paths
# load image fron disk, resize it and convert it to grayscale
print(f'[INFO] processing {file.filename}')
image = cv2.imread(image_path)
image = imutils.resize(image, width=1024)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
clone = image.copy()
# detect faces in grayscale image
rects = detector(gray, 1)
# loop over face detections
for rect in rects:
# determine facial landmarks for face region, then align face
shape = predictor(gray, rect)
face = fa.align(image, gray, rect)
# draw bounding box around face
x, y, w, h = face_utils.rect_to_bb(rect)
cv2.rectangle(clone, (x, y), (x + w, y + h), (0, 255, 0), 2)
if config.DATASET == 'IOG':
# load Label Encoder and mean files
print('[INFO] loading label encoders and mean files...')
age_le = pickle.loads(open(deploy.AGE_LABEL_ENCODER, 'rb').read())
gender_le = pickle.loads(
open(deploy.GENDER_LABEL_ENCODER, 'rb').read())
age_means = json.loads(open(deploy.AGE_MEAN).read())
gender_means = json.loads(open(deploy.GENDER_MEAN).read())
# initialize image preprocessors
age_mp = MeanPreprocessor(age_means['R'], age_means['G'],
age_means['B'])
gender_mp = MeanPreprocessor(gender_means['R'], gender_means['G'],
gender_means['B'])
age_preds, gender_preds = predict(face, sp, age_mp, gender_mp, cp,
iap, deploy.AGE_NETWORK_PATH,
deploy.GENDER_NETWORK_PATH,
age_le, gender_le)
elif config.DATASET == 'ADIENCE':
# age_preds_cross, gender_preds_cross = [], []
i = 0
# load Label Encoder and mean files
print(
f'[INFO] loading label encoders and mean files for cross validation {i}...'
)
age_le = pickle.loads(
open(deploy.AGE_LABEL_ENCODERS[i], 'rb').read())
gender_le = pickle.loads(
open(deploy.GENDER_LABEL_ENCODERS[i], 'rb').read())
age_means = json.loads(open(deploy.AGE_MEANS[i]).read())
gender_means = json.loads(open(deploy.GENDER_MEANS[i]).read())
# initialize image preprocessors
age_mp = MeanPreprocessor(age_means['R'], age_means['G'],
age_means['B'])
gender_mp = MeanPreprocessor(gender_means['R'], gender_means['G'],
gender_means['B'])
age_preds, gender_preds = predict(face, sp, age_mp, gender_mp, cp,
iap, deploy.AGE_NETWORK_PATHS[i],
deploy.GENDER_NETWORK_PATHS[i],
age_le, gender_le)
# age_preds_cross.append(age_pred)
# gender_preds_cross.append(gender_pred)
# age_preds, gender_preds = np.mean(age_preds_cross, axis = 0), np.mean(gender_preds_cross, axis = 0)
clone = AgeGenderHelper.visualize_video(age_preds, gender_preds,
age_le, gender_le, clone,
(x, y))
# path = image_path.split('.')
# pred_path = '.'.join([f'{path[0]}_predict', path[1]])
# pred_filename = pred_path.split(os.path.sep)[-1]
pred_path = '.'.join([f"{image_path.split('.')[0]}_1", 'jpg'])
cv2.imwrite(pred_path, clone)
# image_url = uploader.upload(pred_path)
gc.collect()
K.clear_session()
return render_template('index.html',
filename=pred_path.split(os.path.sep)[-1])
(startX, startY, endX, endY) = box.astype('int')
face_locations.append((startX, startY, endX, endY))
#print(len(face_locations))
if len(face_locations) == 0:
cv2.imshow('face', img)
key = cv2.waitKey(1) & 0xFF
if key == ord("q"):
break
continue
#cv2.destroyAllWindows()
for i in range(len(face_locations)):
(startX, startY, endX, endY) = face_locations[i]
faceAligned = fa.align(img, img, dlib.rectangle(startX, startY, endX, endY))
face_array = faceAligned[:,:,::-1]
#face_array = face_array[startY:endY, startX:endX]
face_encode = face_recognition.face_encodings(face_array)
face_encode = np.array(face_encode)
#print(face_encode.shape)
if face_encode.shape[0] == 0:
cv2.rectangle(img, (startX, startY), (endX, endY),
(0, 0, 255), 2)
#cv2.imshow('face', img)
#key = cv2.waitKey(1) & 0xFF
#if key == ord("q"):
#break
frame = vs.read()
frame = imutils.resize(frame, width=600)
height, width = frame.shape[:2]
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# detect faces in the grayayscale frame
rects = detector(gray_frame, 0)
key = cv2.waitKey(1) & 0xFF
# if the `q` key was pressed, break from the loop
if key == ord("q"):
break
# if the `s` key was pressed save the first found face
if key == ord('s'):
if len(rects) > 0:
faceAligned = fa.align(frame, gray_frame, rects[0])
image_name = base_dir + str(count_image) + ".png"
# save image
cv2.imwrite(image_name, faceAligned)
# show image
cv2.imshow(image_name, faceAligned)
count_image += 1
# if count_image > 10:
# count_image = 0
# loopop over the face detections
for rect in rects:
(x, y, w, h) = face_utils.rect_to_bb(rect)
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 0, 255), 1)
# show the frame
input(" Press enter to continue ! ")
check.clear() # init list
while (True):
ret, frame = cap.read()
faceAlign = frame.copy()
start = time.time()
dets, scores, idx = detector.run(frame, 0, tolerance)
for i, d in enumerate(dets):
cv2.rectangle(frame, (d.left(), d.top()), (d.right(), d.bottom()),
(255, 0, 0), 2) # 將人臉先框起來
faceAlign = fa.align(
frame, frame,
d) # new photo , 由原本的圖片變成只有臉的圖片， photo size = 256 * 256
dets, scores, idx = detector.run(faceAlign, 0,
0) # new photo 還需要再 detect一次
for i1, d1 in enumerate(dets):
shape = sp(faceAlign, d1) # mark 68_face_landmarks
face_descriptor = np.array(
[facerec.compute_face_descriptor(faceAlign, shape)])
prediction = cnn_model.predict_proba(face_descriptor)
highest_proba = 0
counter = 0
# print prediction
for prob in prediction[0]:
if prob > highest_proba and prob >= 0.1:
highest_proba = prob
def test():
# 取得dlib預設的臉部偵測器
detector = dlib.get_frontal_face_detector()
# 根據shape_predictor方法載入68個特徵點模型，此方法為人臉表情識別的偵測器
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
# 臉部校正
fa = FaceAligner(predictor, desiredFaceWidth=256)
# 載入人臉辨識檢測器
facerec = dlib.face_recognition_model_v1(
"dlib_face_recognition_resnet_model_v1.dat")
# 比對人臉描述子列表
descriptors = []
# 比對人臉名稱列表
candidate = []
# 圖片路徑
path = []
# 開鏡頭前先播放聲音
playsound('./sound/voice1_ch.mp3')
playsound('./sound/voice1_en.mp3')
playsound('./sound/voice1_j.mp3')
# Play Music on Separate Thread (in background)
# music_thread = Thread(target=play_music)
# music_thread.start()
# 開啟影片檔案
cap = cv2.VideoCapture(0)
# 比對人臉圖片資料夾名稱
faces_folder_path = "./rec"
WHO = None
Erro = 0
# 有讀過的照片存成新的檔可直接使用，有新的圖片再加
# 路徑中的file個數
NUM_OF_FILES = 0 # 檔案數
for fn in os.listdir(faces_folder_path):
NUM_OF_FILES += 1
#print(NUM_OF_FILES)
person = [0 for i in range(NUM_OF_FILES)]
#print(person)
# 原本的個數
f = open("numOfFiles.txt")
num_o = f.read()
num_o = int(num_o)
h5file = './desc.h5'
#print(sorted(glob.glob(os.path.join(faces_folder_path, "*.jpg")),key=os.path.getmtime))
# 所有檔名
for f in sorted(glob.glob(os.path.join(faces_folder_path, "*.jpg")),
key=os.path.getmtime):
base = os.path.basename(f)
# print(base)
# 依序取得圖片檔案人名，存到candidate一維陣列中
candidate.append(os.path.splitext(base)[0])
# print(candidate)
path.append(f)
#print(path)
# 如果新的檔案數與舊的檔案數不同
if num_o != NUM_OF_FILES:
# 把新的個數寫到txt檔裡
num = open('numOfFiles.txt', 'w')
num.write(str(NUM_OF_FILES))
num.close()
# 新增檔名 new NUM_OF_FILES ori num_o
c = open("cand.txt", "a", encoding="utf-8")
for i in range(num_o - 1, NUM_OF_FILES - 1, 1): # -2 -1是mac檔多一個檔案
c.write('\n')
c.write(candidate[i])
c.close()
# 新增one shot計算匡
# 讀取資料夾裡的圖片及檔案名(人名)，並將每張圖的128維特徵向量存到description一維陣列中
h5f = h5py.File(h5file, 'a')
for i in range(num_o - 1, NUM_OF_FILES - 1, 1):
img = io.imread(path[i])
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
# 1.人臉偵測
face_rects = detector(img, 0)
# print(face_rects)
for index, face in enumerate(face_rects):
# 2.人臉校正
faceAligned = fa.align(img, gray, face)
# 3.再人臉偵測(去除校正後多餘的部分)
face_rects2 = detector(faceAligned, 1)
for index2, face2 in enumerate(face_rects2):
ax1 = face2.left()
ay1 = face2.top()
ax2 = face2.right()
ay2 = face2.bottom()
# 4.68特徵點偵測
shape = predictor(faceAligned, face2)
# 5.取得描述子，128維特徵向量
face_descriptor = facerec.compute_face_descriptor(
faceAligned, shape)
# 轉換numpy array格式
v = np.array(face_descriptor)
descriptors.append(v)
#print(face_rects)
# original的在face_allIn_final0.py
h5f.create_dataset('new{}'.format(i), data=v)
h5f.close()
# 存新的到原本的txt檔
np.savetxt('desc_txt.txt', descriptors)
# 檢查用
# 讀檔
kk = h5py.File(h5file, 'r')
#print(type(h5f))
#print(h5f)
# 通過切片得到numpy陣列
#print(list(kk.keys()))
#print(h5f['new8'][:])
#下面這條要在裡面執行，現在在外面
#for i in range(0, NUM_OF_FILES-1, 1):
#print(kk['new{}'.format(i)][:])
# 以迴圈從影片檔案讀取影格，並顯示出來
while cap.isOpened():
k = open("cand.txt", encoding="utf-8")
cand = k.readlines()
x1 = 0
x2 = 0
y1 = 0
y2 = 0
# 從視訊鏡頭擷取畫面
ret, frame = cap.read()
# 縮小圖片
frame = imutils.resize(frame, width=800) # 顯示影像框框
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
# 1.人臉偵測
face_rects = detector(frame, 1)
# 取出所有偵測的結果(所有人臉座標點)
for index, rect in enumerate(face_rects):
x1 = rect.left()
y1 = rect.top()
x2 = rect.right()
y2 = rect.bottom()
# 以方框標示偵測的人臉
cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 4,
cv2.LINE_AA)
# 2.人臉校正
faceAligned = fa.align(frame, gray, rect)
# 3.再人臉偵測(去除校正後多餘的部分)
face_rects2 = detector(faceAligned, 1)
# 取出所有偵測的結果(所有人臉座標點)
for index2, rect2 in enumerate(face_rects2):
ax1 = rect2.left()
ay1 = rect2.top()
ax2 = rect2.right()
ay2 = rect2.bottom()
# 4.68特徵點偵測
shape = predictor(faceAligned, rect2)
# 5.取得描述子，128維特徵向量
face_descriptor = facerec.compute_face_descriptor(
faceAligned, shape)
# 轉換numpy array格式
d_test = np.array(face_descriptor)
# 計算歐式距離 (與圖片庫裡，各個人臉間的距離)(5張照片就有5個距離)
# 開h5檔案
h5f = h5py.File(h5file, 'r')
# 清空 存放人臉距離的陣列
dist = []
for i in range(0, NUM_OF_FILES - 1, 1): #-1就好
# 計算距離
dist_ = np.linalg.norm(h5f['new{}'.format(i)][:] - d_test)
# 加入陣列
dist.append(dist_)
# 辨識人名
if dist != []:
# 將比對人名和比對出來的歐式距離組成一個dict
c_d = dict(zip(cand, dist))
# 根據歐式距離由小到大排序 [("名字",距離)]二微陣列
cd_sorted = sorted(c_d.items(), key=lambda d: d[1])
#print(cd_sorted)
# 歐式距離(0~1)越小越像，設定0.5作為最低辨識標準
if cd_sorted[0][1] < 0.4:
rec_name = cd_sorted[0][0]
else:
rec_name = "No Data"
Erro = Erro + 1
if Erro == 6:
WHO = "No Data"
cap.release()
cv2.destroyAllWindows()
return WHO
break
# 標示辨識的人名(中文)
imgPil = Image.fromarray(frame)
font = ImageFont.truetype("C:/Windows/Fonts/msjh.ttc", 20)
draw = ImageDraw.Draw(imgPil)
draw.fontmode = '1' # 關閉反鋸齒
draw.text((x1, y1 - 20),
rec_name,
font=font,
fill=(255, 255, 255))
frame = np.array(imgPil)
for index, name in enumerate(cand):
if rec_name == name:
person[index] = person[index] + 1
# print(person)
if person[index] == 4:
WHO = cand[index]
# 標示辨識的人名(中文)
if WHO is not None:
time.sleep(5)
cap.release()
cv2.destroyAllWindows()
print(WHO)
return WHO
break
elif sum(person) > 8:
WHO = "No Data"
cap.release()
cv2.destroyAllWindows()
return WHO
break
# 標示辨識的人名(只能標示英文)
# cv2.putText(frame, rec_name, (x1, y1), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2, cv2.LINE_AA)
# CV2是用BGR
frame = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
# 顯示結果
cv2.imshow("Face Detection", frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
def Euclied_dist(inp1, inp2):
return np.sqrt(np.sum(np.square(inp1 - inp2)))
images = []
for i in os.listdir(base_dir):
for j in os.listdir(os.path.join(base_dir, i)):
img = cv2.imread(os.path.join(base_dir, i, j))
#print(i,j)
faces = fd(img, 1)
f = faces[0]
#print(len(faces))
x, y, w, h = f.left(), f.top(), (f.right() - f.left()), (f.bottom() -
f.top())
temp = fa.align(img, cv2.cvtColor(img, cv2.COLOR_BGR2GRAY), f)
temp = cv2.resize(cv2.cvtColor(temp, cv2.COLOR_BGR2GRAY), (160, 160))
cv2.imwrite(os.path.join(base_dir, r'temp.jpg'), temp)
temp = cv2.imread(os.path.join(base_dir, r'temp.jpg'))
os.remove(os.path.join(base_dir, r'temp.jpg'))
images.append(temp)
images = Standardize(np.array(images))
np.save(os.path.join(r'D:\ProjectML\FR', r'FImages'), images)
images = np.load(os.path.join(r'D:\ProjectML\FR', r'FImages.npy'))
# LABEL ENCODING FOR ALL THE FACES i.E CONVERTING NAMES OF FACES TO NUMERICAL VALUE
labels = []
for i in os.listdir(base_dir):
args = vars(ap.parse_args())
# initialize dlib's face detector (HOG-based) and then create
# the facial landmark predictor and the face aligner
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(args["shape_predictor"])
fa = FaceAligner(predictor, desiredFaceWidth=256)
# load the input image, resize it, and convert it to grayscale
image = cv2.imread(args["image"])
image = imutils.resize(image, width=800)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# show the original input image and detect faces in the grayscale
# image
cv2.imshow("Input", image)
rects = detector(gray, 2)
# loop over the face detections
for rect in rects:
# extract the ROI of the *original* face, then align the face
# using facial landmarks
(x, y, w, h) = rect_to_bb(rect)
faceOrig = imutils.resize(image[y:y + h, x:x + w], width=256)
faceAligned = fa.align(image, gray, rect)
# display the output images
cv2.imshow("Original", faceOrig)
cv2.imshow("Aligned", faceAligned)
cv2.waitKey(0)
def detect_faces(color_image_list, gray_image_list, dlib_models):
"""
Detects faces using Dlib's CNN model.
Args:
color_image_list (list): list of images. Each item is a frame read by the cv2 package.
gray_image_list (list): list of images in grayscale. This list should contain the same images
as the color_image_list, but in grayscale. This list is used by the CNN model.
dlib_models (dict): a dictionary containing dlib cnn, shape_predictor, and recognition models.
Returns:
face_images (np.array): an array of images of detected faces.
n_faces_list (list): a list of ints showing the number of detected faces in each frame.
flat_face_rects (list): a list of dlib.rectangle objects containing rectangle info of each detected face.
face_descriptors (list): list of face_descriptor. A face_descriptor is a lists of 128 dim vector that describes the face.
Example:
Given the following inputs
color_image_list = [Img1 (has three faces), Img2 (has two faces)]
gray_image_list = [Img1_grayscale (has three faces), Img2_grayscale (has two faces)]
This function produces the following
face_images = np.array([face1_img1, face2_img1, face3_img1, face4_img2, face5_img2])
n_faces_list = [3, 2]
flat_face_rects = [r1, r2, r3, r4, r5]
Note that the face images are cropped from the original image in the arguments. So, in our
example, face1_img1 has a smaller size of Img1, and face2_img1 might have a compeletely
different size but again smaller than Img1.
"""
mmod_rects = dlib_models['cnn'](gray_image_list, upsample_num_times=UPSAMPLE_COUNT)
flat_face_rects = []
flat_image_list_indices = []
n_faces_list = []
all_shapes_list = []
# mmod_rects is a list of list of rectangles
for i, image_detection_rects in enumerate(mmod_rects):
rects = dlib.rectangles()
# save rects into an array to use later
rects.extend([d.rect for d in image_detection_rects])
flat_face_rects.extend(rects)
flat_image_list_indices.extend([i]*len(image_detection_rects))
n_faces_list.append(len(image_detection_rects))
# find shapes in the image -- this is used for face recognition
faces = dlib.full_object_detections()
for r in rects:
shape = dlib_models['shape_predictor'](color_image_list[i], r)
faces.append(shape)
all_shapes_list.append(faces)
# in the above example
# flat_face_rects = [r1, r2, r3, r4, r5]
# flat_image_list_indices = [0, 0, 0, 1, 1]
# n_faces_list = [3, 2]
# all_shapes_list = [dlib.full_object_detections, dlib.full_object_detections]
# align detected rectangles to get faces for the next step
fa = FaceAligner(dlib_models['shape_predictor'])
face_images = []
for i, rect in enumerate(flat_face_rects):
image_index = flat_image_list_indices[i]
aligned_image = fa.align(color_image_list[image_index], gray_image_list[image_index], rect)
aligned_image = imutils.resize(aligned_image, width=160, height=160)
face_images.append(aligned_image)
# in the above example
# face_images = [img1, img2, img3, img4, img5]
# face encodings
face_descriptors = dlib_models['recognition_model'].compute_face_descriptor(color_image_list, all_shapes_list)
# face_descriptor is a lists of 128 dim vector that describes the face.
# if two face descriptor vectors have a Euclidean distance between them less than 0.6
# then they are from the same person
# in the above example
# face_descriptors = [[[0..127],[0..127],[0..127]],[[0..127],[0..127]]]
return np.array(face_images), n_faces_list, flat_face_rects, face_descriptors
def mark_your_attendance(request):
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(
'face_recognition_data/shape_predictor_68_face_landmarks.dat'
) #Add path to the shape predictor ######CHANGE TO RELATIVE PATH LATER
svc_save_path = "face_recognition_data/svc.sav"
with open(svc_save_path, 'rb') as f:
svc = pickle.load(f)
fa = FaceAligner(predictor, desiredFaceWidth=96)
encoder = LabelEncoder()
encoder.classes_ = np.load('face_recognition_data/classes.npy')
faces_encodings = np.zeros((1, 128))
no_of_faces = len(svc.predict_proba(faces_encodings)[0])
count = dict()
present = dict()
log_time = dict()
start = dict()
for i in range(no_of_faces):
count[encoder.inverse_transform([i])[0]] = 0
present[encoder.inverse_transform([i])[0]] = False
vs = VideoStream(src=0).start()
sampleNum = 0
while (True):
frame = vs.read()
frame = imutils.resize(frame, width=800)
gray_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
faces = detector(gray_frame, 0)
for face in faces:
print("INFO : inside for loop")
(x, y, w, h) = face_utils.rect_to_bb(face)
face_aligned = fa.align(frame, gray_frame, face)
cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 1)
(pred, prob) = predict(face_aligned, svc)
if (pred != [-1]):
person_name = encoder.inverse_transform(np.ravel([pred]))[0]
pred = person_name
if count[pred] == 0:
start[pred] = time.time()
count[pred] = count.get(pred, 0) + 1
if count[pred] == 4 and (time.time() - start[pred]) > 1.2:
count[pred] = 0
else:
#if count[pred] == 4 and (time.time()-start) <= 1.5:
present[pred] = True
log_time[pred] = datetime.datetime.now()
count[pred] = count.get(pred, 0) + 1
print(pred, present[pred], count[pred])
cv2.putText(frame,
str(person_name) + str(prob), (x + 6, y + h - 6),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1)
else:
person_name = "unknown"
cv2.putText(frame, str(person_name), (x + 6, y + h - 6),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 1)
#cv2.putText()
# Before continuing to the next loop, I want to give it a little pause
# waitKey of 100 millisecond
#cv2.waitKey(50)
#Showing the image in another window
#Creates a window with window name "Face" and with the image img
cv2.imshow("Mark Attendance - In - Press q to exit", frame)
#Before closing it we need to give a wait command, otherwise the open cv wont work
# @params with the millisecond of delay 1
#cv2.waitKey(1)
#To get out of the loop
key = cv2.waitKey(50) & 0xFF
if (key == ord("q")):
break
#Stoping the videostream
vs.stop()
# destroying all the windows
cv2.destroyAllWindows()
update_attendance_in_db_in(present)
# return redirect('recognition/admin-dashboard')
return render(request, 'recognition/admin_dashboard.html')
cv2.imshow("Greška", image)
cv2.waitKey(30)
#čekanje da korisnik primjeti sliku
time.sleep(1.2)
cv2.destroyAllWindows(
) # uništavanje prozora slike i nastavak rada petlje
continue
#ako slika ima samo jedno lice prikazat je
cv2.imshow("Orginal slika", image)
for rect in rects:
#pretvaranje pravougaonika lica u visini širinu i početne tačke x,y
(x, y, w, h) = rect_to_bb(rect)
#smanjivanje pronađenog lica na 256 piksela širine
faceOrig = imutils.resize(image[y:y + h, x:x + w], width=256)
faceAligned = fa.align(image, gray, rect) #rotiranje lica
LokacijaSlike = str(
pathlib.Path(__file__).parent.absolute().joinpath("podaci").
joinpath("korisnici").joinpath(folder).joinpath(slike))
height, witdh, channell = faceAligned.shape
CentarSlike = (w / 2, h / 2)
#preokret lica za 180 stepeni
Rotirana = imutils.rotate(faceAligned, 180)
cv2.imwrite(LokacijaSlike, Rotirana)
cv2.imshow("Lice Neispravljeno", faceOrig)
cv2.imshow("Lice Ispravljeno", Rotirana)
cv2.waitKey(30)
time.sleep(1.2)
# detect faces in the grayscale image
cv2.imshow("Input", image)
rects = detector(gray, 1)
# loop over the faces that are detected
for (i, rect) in enumerate(rects):
# Detected face landmark (x, y)-coordinates are converted into
# Numpy array
shape = predictor(gray, rect)
shape = face_utils.shape_to_np(shape)
# convert dlib's rectangle to OpenCV bounding box and draw
# [i.e., (x, y, w, h)]
(x, y, w, h) = face_utils.rect_to_bb(rect)
faceOrig = imutils.resize(image[y:y + h, x:x + w], width=256)
faceAligned = face.align(image, gray, rect)
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
import uuid
f = str(uuid.uuid4())
cv2.imwrite("foo/" + f + ".png", faceAligned)
# shows the face number
cv2.putText(image, "Face #{}".format(i + 1), (x - 10, y - 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
# loop over the (x, y) coordinate for facial landmark and drow on th image
for (x, y) in shape:
cv2.circle(image, (x, y), 1, (0, 0, 255), -1)
cv2.imshow("Original", faceOrig)
(0, 0, 255), 2)
# Draw a label with a name below the face
cv2.rectangle(frame, (left, bottom - 35),
(right, bottom), (0, 0, 255),
cv2.FILLED)
font = cv2.FONT_HERSHEY_DUPLEX
cv2.putText(frame, name, (left + 6, bottom - 6),
font, 1.0, (255, 255, 255), 1)
# Estimate gender and age
# Convert to rect
face_detected = _css_to_rect(
(top, right, bottom, left))
print(face_detected)
faces[0, :, :, :] = fa.align(
rgb_small_frame, frame_gray, face_detected)
age_predict, gender_predict = sess.run(
[age, gender],
feed_dict={
images_pl: faces,
train_mode: False
})
label = "{}, {}".format(
int(age_predict),
"F" if gender_predict == 0 else "M")
cv2.putText(frame, label, (left + 6, top - 6),
font, 1.0, (255, 255, 255), 1)
c.execute(
"insert into people_inroom values (%s, %s, %s)",