def test():
detector = dlib.simple_object_detector("../boobs.svm")
win_det = dlib.image_window()
win_det.set_image(detector)
# dlib.hit_enter_to_continue()
print("Showing detections on the images in the faces folder...")
win = dlib.image_window()
tp = 0
fn = 0
dur = 0
test_path = "../pics/test/*.jpg"
# test_path = "/home/external/moderation-p**n-detector/boobs-oboobs/*.jpg"
for f in glob.glob(test_path):
try:
img = io.imread(f)
except IOError as e:
print("Image {} can't be loaded: {}".format(f, e.message))
continue
try:
t_start = time.clock()
dets = detector(img)
t_end = time.clock()
except RuntimeError as e:
print("Image {} can't be detected: {}".format(f, e.message))
continue
dur += t_end - t_start
num = len(dets)
if num > 0:
print("Boobs {} detected in file: {}".format(num, f))
for k, d in enumerate(dets):
print(
"Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()
)
)
win.clear_overlay()
win.set_image(img)
win.add_overlay(dets)
# dlib.hit_enter_to_continue()
tp += 1
else:
fn += 1
if tp + fn > 100:
break
print("tp={} fn={} precision={} dur={}".format(tp, fn, 1.0 * tp / (tp + fn), 1.0 * dur / (tp + fn)))
def __init__(self, visulize = True):
self.detector = dlib.get_frontal_face_detector()
self.predictor = dlib.shape_predictor(predictor_path)
self.fronter = getDefaultFrontalizer()
if visulize:
self.win = dlib.image_window()
self.win2 = dlib.image_window()
else:
self.win = self.win2 = None
def classify(img):
detector = dlib.simple_object_detector("cupdetector_2.svm")
win_det = dlib.image_window()
win_det.set_image(detector)
win = dlib.image_window()
test_dir = '/home/jyotiska/Dropbox/Computer Vision/Cups_test'
convert_dir = '/home/jyotiska/Dropbox/Computer Vision/Cups_test_convert'
assorted_dir = '/home/jyotiska/Dropbox/Computer Vision/Item bucket'
items = os.listdir(assorted_dir)
convert_i = 0
for f in glob.glob(convert_dir+"/*.*"):
print "processing file:", f
img = io.imread(f)
extension = f.split(".")[1]
convert_file = "convert_"+str(convert_i)+"."+extension
shutil.copy(f,convert_file)
print "convert file:",convert_file
background = Image.open(convert_file)
dets = detector(img)
print "number of cups detected:", len(dets)
for d in dets:
x = d.left()
y = d.top()
width = d.right() - x
height = d.bottom() - y
print " detection position left,top,right,bottom:", d.left(), d.top(), d.right(), d.bottom()
print width,height
r = random.randint(0,len(items)-1)
print r,items[r]
random_item = Image.open(assorted_dir+"/"+items[r])
# scale it a bit more, and adjust position
# Apply blur?
resized = random_item.resize( (int(1.2*width),int(1.2*height)) )
background.paste(resized, (d.left()-12,d.top()-10), resized)
background.show()
background.save(convert_file)
win.clear_overlay()
win.set_image(img)
win.add_overlay(dets)
convert_i += 1
def camera_fd():
import cv2
cam = cv2.VideoCapture(-1)
win = dlib.image_window()
detector = detector = dlib.get_frontal_face_detector()
def detect():
s, img = cam.read()
if not s:
return
img2 = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
dets = detector(img2)
num = len(dets)
if num > 0:
print("found!")
win.clear_overlay()
win.set_image(img2)
win.add_overlay(dets)
try:
while True:
detect()
except KeyboardInterrupt as e:
print("exiting")
cam.release()
cv2.destroyAllWindows()
def camera_boobs():
import cv2
# cam = cv2.VideoCapture(-1)
cam = cv2.VideoCapture("/home/nick/temp/boobs/b1.flv")
win = dlib.image_window()
detector = dlib.simple_object_detector("detector.svm")
def detect():
s, img = cam.read()
if not s:
return
img2 = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
dets = detector(img2)
num = len(dets)
if num > 0:
print("found!")
win.clear_overlay()
win.set_image(img2)
win.add_overlay(dets)
try:
while True:
detect()
except KeyboardInterrupt as e:
print("exiting")
cam.release()
cv2.destroyAllWindows()
def main():
if len(sys.argv) < 3:
print(
"Give the path to the examples/faces directory as the argument to this "
"program. For example, if you are in the python_examples folder then "
"execute this program by running:\n"
" ./detect.py ../examples/faces ./detector.svm [-s]")
exit()
f = sys.argv[1]
detector_fp = sys.argv[2]
s = False
if len(sys.argv) > 3 and sys.argv[3] == '-s':
s = True
dets = handle_locations(f, detector_fp)
print("Number of detected: {}".format(len(dets)))
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
if s:
win = dlib.image_window()
win.clear_overlay()
win.set_image(img)
win.add_overlay(dets)
dlib.hit_enter_to_continue()
def __init__(self):
self._detector = dlib.get_frontal_face_detector()
pose_predictor_path = '../models/dlib/shape_predictor_68_face_landmarks.dat'
self._predictor = dlib.shape_predictor(pose_predictor_path)
if not self.SERVER_NO_GUI_MODE:
self._win = dlib.image_window()
self._lock = RLock()
def show_pair(path1, path2):
global win1, win2
if win1 == None:
win1 = dlib.image_window()
if win2 == None:
win2 = dlib.image_window()
img1 = io.imread(path1)
img2 = io.imread(path2)
win1.clear_overlay()
win2.clear_overlay()
win1.set_image(img1)
win2.set_image(img2)
dlib.hit_enter_to_continue()
win1.set_image(img1)
win2.set_image(img2)
def display_landmarks(img, dets, shapes):
win = dlib.image_window()
win.clear_overlay()
win.set_image(img)
for shape in shapes:
win.add_overlay(shape)
win.add_overlay(dets)
dlib.hit_enter_to_continue()
def main():
detector = LandmarkDetector()
img = cv2.imread("../data/test_data/multi_processed.jpg")
shape = detector.get_landmarks(img)
win = dlib.image_window()
win.set_image(img)
win.add_overlay(shape)
print("Part 0: {}, Part 1: {} ...".format(shape.part(0),
shape.part(1)))
def show_learned_hog_filter(self):
# Now let's use the detector as you would in a normal application. First we
# will load it from disk.
#print "Loading detector"
#detector = dlib.simple_object_detector("../data/models/detector.svm")
# We can look at the HOG filter we learned. It should look like a face. Neat!
win_det = dlib.image_window()
print("MTB filter")
win_det.set_image(self.detectors[0])
dlib.hit_enter_to_continue()
print("PED filter")
win_det.set_image(self.detectors[1])
dlib.hit_enter_to_continue()
def DetectFaceLandmarksInList(frameList, faceDetector = None, shapePredictor = None, skipLength = 2, debug = False):
'''
Given a frame list, detect (track) the faces
Returns details of the facial landmarks for each frame
'''
if ((faceDetector is None) or (shapePredictor is None)):
predictorPath = 'coreData/shape_predictor_68_face_landmarks.dat'
faceDetector = dlib.get_frontal_face_detector()
shapePredictor = dlib.shape_predictor(predictorPath)
if (debug):
win = dlib.image_window()
win.clear_overlay()
faceList = []
for i in range(0, frameList.shape[0], skipLength):
frame = frameList[i]
dets = faceDetector(frame, 1)
if debug:
win.clear_overlay()
win.set_image(frame)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(k, d.left(), d.top(), d.right(), d.bottom()))
shape = shapePredictor(frame, d)
win.add_overlay(shape)
win.add_overlay(dets)
dets = list(enumerate(dets))
faceNum = len(dets)
if (faceNum == 1):
shape = shapePredictor(frame, dets[0][1])
# faceShape = NormalizeShape(shape, dets[0][1])
faceShape = RawShape(shape, dets[0][1])
faceList.append(faceShape)
faceList = np.array(faceList)
return faceList
def run(predictor_path, faces_folder_path):
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_path)
win = dlib.image_window()
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
# Ask the detector to find the bounding boxes of each face. The 1 in the
# second argument indicates that we should upsample the image 1 time. This
# will make everything bigger and allow us to detect more faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
# Get the landmarks/parts for the face in box d.
shape = predictor(img, d)
print("Part 0: {}, Part 1: {} ...".format(shape.part(0),
shape.part(1)))
left_eye_idxs = range(42, 48)
left_eye_pts = np.array([[p.x, p.y] for i, p in enumerate(shape.parts()) if i in left_eye_idxs])
left_eye = np.mean(left_eye_pts, axis=0)
right_eye_idxs = range(36, 42)
right_eye_pts = np.array([[p.x, p.y] for i, p in enumerate(shape.parts()) if i in right_eye_idxs])
right_eye = np.mean(right_eye_pts, axis=0)
cv2.circle(img, tuple(left_eye.astype(np.int)), 3, color=(0, 255, 255))
cv2.circle(img, tuple(right_eye.astype(np.int)), 3, color=(0, 255, 255))
# Draw the face landmarks on the screen.
win.add_overlay(shape)
win.set_image(img)
win.add_overlay(dets)
cv2.imwrite(os.path.join(face_folder_path, f.replace('.', '_landmarks.')), img)
dlib.hit_enter_to_continue()
win.clear_overlay()
def __init__(self):
#initializes frame
self.frame = None
#initializes ros node for face detect, pubilishes to face location
rospy.init_node('face_detect', anonymous=True)
self.pub = rospy.Publisher('/face_location', String, queue_size=10)
#definines file paths
rospack = rospkg.RosPack()
PACKAGE_PATH = rospack.get_path("edwin")
self.predictor_path = PACKAGE_PATH + '/params/shape_predictor_68_face_landmarks.dat'
self.faces_folder_path = PACKAGE_PATH + '/params'
#def of attributes
self.detect = True
self.detector = dlib.get_frontal_face_detector()
self.predictor = dlib.shape_predictor(self.predictor_path)
self.window = dlib.image_window()
#CvBridge to usb_cam, subscribes to usb cam
self.bridge = CvBridge()
rospy.Subscriber("usb_cam/image_raw", Image, self.img_callback)
def main():
videofile = sys.argv[1]
print('[INFO] Reading %s' % videofile )
vid = imageio.get_reader( videofile )
nFrames = 0
frames = []
for i in range(1000, 3000):
img = vid.get_data( i )
frames.append( img )
nFrames += 1
print('[INFO] Loaded all frames' )
bbox_ = get_rectangle( frames[0] )
# Create the correlation tracker - the object needs to be initialized
# before it can be used
tracker = dlib.correlation_tracker()
win = dlib.image_window()
# We will track the frames as we load them off of disk
for k, img in enumerate(frames):
print("Processing Frame {}".format(k))
# We need to initialize the tracker on the first frame
if k == 0:
# Start a track on the juice box. If you look at the first frame you
# will see that the juice box is contained within the bounding
# box (74, 67, 112, 153).
(x0, y0), (x1, y1) = bbox_
tracker.start_track(img, dlib.rectangle( x0, y0, x1, y1 ))
else:
# Else we just attempt to track from the previous frame
tracker.update(img)
win.clear_overlay()
win.set_image(img)
win.add_overlay(tracker.get_position())
def onVideo(self):
motionChange = False
predictor_path = "/home/lee/Software/test/shape_predictor_68_face_landmarks.dat"
face_rec_model_path = "/home/lee/Software/test/dlib_face_recognition_resnet_model_v1.dat"
faces_folder_path = "/home/lee/Software/pycharm_test/candidate-faces"
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_path)
facerec = dlib.face_recognition_model_v1(face_rec_model_path)
descriptors = []
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
# win.clear_overlay()
# win.set_image(img)
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
shape = predictor(img, d)
# win.clear_overlay()
# win.add_overlay(d)
# win.add_overlay(shape)
face_descriptor = facerec.compute_face_descriptor(img, shape)
# print ("face_descriptor: {}".format(face_descriptor))
v = numpy.array(face_descriptor)
# print ("v: {}".format(v))
descriptors.append(v)
# print ("descriptors: {}".format(descriptors))
win = dlib.image_window()
cap = cv2.VideoCapture(0)
c = 1
time = 10
while cap.isOpened():
ret, cv_img = cap.read()
if cv_img is None:
break
image = cv2.cvtColor(cv_img, cv2.COLOR_RGB2BGR)
win.clear_overlay()
win.set_image(image)
dets = detector(image, 0)
dist = []
# print("Number of faces detected: {}".format(len(dets)))
# lists = [[0 for k in range(4)] for j in range(10)]
if (c % time == 0):
for k, d in enumerate(dets):
# print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
# i, d.left(), d.top(), d.right(), d.bottom()))
shape1 = predictor(image, d)
# print("Part 0: {}, Part 1: {}, Part 2: {} ...".format(shape.part(0), shape.part(1), shape.part(2)))
win.clear_overlay()
win.add_overlay(d)
win.add_overlay(shape1)
face_descriptor1 = facerec.compute_face_descriptor(
image, shape1)
# print ("face_descriptor1: {}".format(face_descriptor1))
d_test = numpy.array(face_descriptor1)
# print ("d_test: {}".format(d_test))
for i in descriptors:
dist_ = numpy.linalg.norm(i - d_test)
dist.append(dist_)
# print("dist: {}".format(dist))
candidate = [
'张丽梅', '丁高兴', '徐科杰', '金莹莹', '李政英', '石光耀', '陈美利', '段宇乐',
'李宗辉', '陈杰'
]
c_d = dict(zip(candidate, dist))
cd_sorted = sorted(c_d.iteritems(), key=lambda d: d[1])
# print (cd_sorted)
if (((cd_sorted[1][1] - cd_sorted[0][1]) < 0.1) &
(cd_sorted[0][1] > 0.381)):
print "\n 这个人是：陌生人"
j = -1
x = -1
else:
print "\n 这个人是：", cd_sorted[0][0]
if (cd_sorted[0][0] == '李政英'):
j = d.bottom() - d.top()
x = d.left()
elif (cd_sorted[0][0] == '陈美利'):
j = d.bottom() - d.top()
x = d.left()
else:
j = -1
x = -1
if len(dets) == 0:
self.callbackKeyLeft = False
self.callbackKeyRight = False
self.callbackKeyDown = False
self.callbackKeyUp = False
motionChange = True
else:
if x > 300:
self.callbackKeyRight = True
motionChange = True
elif 0 < x < 180:
self.callbackKeyLeft = True
motionChange = True
elif 180 < x < 300:
self.callbackKeyLeft = False
self.callbackKeyRight = False
motionChange = True
if j > 140:
self.callbackKeyDown = True
motionChange = True
elif 0 < j < 110:
self.callbackKeyUp = True
motionChange = True
elif 110 < j < 140:
self.callbackKeyDown = False
self.callbackKeyUp = False
motionChange = True
if motionChange == True:
velocity = 0
velocity += VELOCITYCHANGE if self.callbackKeyUp is True else 0
velocity -= VELOCITYCHANGE if self.callbackKeyDown is True else 0
rotation = 0
rotation += ROTATIONCHANGE if self.callbackKeyLeft is True else 0
rotation -= ROTATIONCHANGE if self.callbackKeyRight is True else 0
# compute left and right wheel velocities
vr = velocity + (rotation / 2)
vl = velocity - (rotation / 2)
# create drive command
cmd = struct.pack(">Bhh", 145, vr, vl)
if cmd != self.callbackKeyLastDriveCommand:
self.sendCommandRaw(cmd)
self.callbackKeyLastDriveCommand = cmd
c = c + 1
# win.clear_overlay()
# win.set_image(img)
# win.add_overlay(dets)
cap.release()
def display_image(img, shape):
win = dlib.image_window()
win.clear_overlay()
win.set_image(img)
if shape:
win.add_overlay(shape)
def view_object_detector(self):
detector = dlib.simple_object_detector(DETECTOR_SVM)
win_det = dlib.image_window()
win_det.set_image(detector)
dlib.hit_enter_to_continue()
def search(img_path, labels_list, threshold=0.5, answer_pic=False):
# if state not in {'recognition', 'search'}:
# raise ValueError('{} not valid'.format(state))
# 候选人脸文件夹
# face_folder_path
# if state == 'recognition':
# 1.加载正脸检测器
detector = dlib.get_frontal_face_detector()
# 2.加载人脸关键点检测器 shape_predictor_68_face_landmarks.dat
sp = dlib.shape_predictor(
r"..\resources\shape_predictor_68_face_landmarks.dat")
# 3. 加载人脸识别模型
facerec = dlib.face_recognition_model_v1(
r"..\resources\dlib_face_recognition_resnet_model_v1.dat")
# image_data = io.imread(image_path)
win = dlib.image_window()
descriptors = []
# 只接受jpg文件
for f in labels_list:
path = '../default_search_labels/{}.jpg'.format(f)
# print("Processing file:{}".format(path))
# 加载标签图片
try:
img = io.imread(path)
except FileNotFoundError as e:
print('标签类型不是jpg,更换为png')
path = '../default_search_labels/{}.png'.format(f)
img = io.imread(path)
# The 1 in the second argument indicates that we should upsample the image
# 1 time. This will make everything bigger and allow us to detect more
# faces.
# 检测人脸数
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
for k, d, in enumerate(dets):
# 人脸关键点检测器sp
shape = sp(img, d)
# 画出人脸区域和关键点
win.clear_overlay()
win.add_overlay(d)
win.add_overlay(shape)
# 3.描述子提取，128D向量
face_descriptor = facerec.compute_face_descriptor(img, shape)
# 转换为numpy array
v = numpy.array(face_descriptor)
descriptors.append(v)
# print(descriptors)
# 加载需测的图片
img = io.imread(img_path)
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
# 为len(dets)张人脸各建一个列表
dist = []
for k, d, in enumerate(dets):
dist_all = []
shape = sp(img, d)
face_descriptor = facerec.compute_face_descriptor(img, shape)
d_test = numpy.array(face_descriptor)
# 计算欧式距离
for i in descriptors:
dist_ = numpy.linalg.norm(i - d_test)
dist_all.append(dist_) # 第k张人脸与第i个标签的比较数据存入列表
dist.append(dist_all)
print(dist)
# 标签库加工，把参数标签文件夹路径中的几个jpg的人名拿出来组成标签列表
print(1)
# list_picture_labels = os.listdir(test_folder_path)
# candidate = []
# for i in list_picture_labels:
# if i[-4:] in ['.jpg', '.JPG', '.PNG', '.png']:
# candidate.append(i[:-4])
# else:
# mb.showwarning('warning', "file invalid!")
# candidate=['xinyuanjieyi','qiaobenhuannai','shiyuanlimei','fengtimo']
candidate = labels_list
c_d = [{}] * len(dets)
answer = []
print(dist)
for i, d, in enumerate(dets):
# c_d[i]记录第i张人脸的标签与可能性值的信息
c_d[i] = dict(zip(candidate, dist[i]))
cd_sorted = sorted(c_d[i].items(), key=lambda d: d[1])
print(cd_sorted)
# 返回信息
# 若dist小于0.4则识别成功，大于则查无此人
# print(cd_sorted[0][1])
# print(threshold)
if cd_sorted[0][1] < threshold:
answer.append(cd_sorted[0][0]) # 查到的人名
# print(d)
# print(type(d))
# 使用opencv在原图上画出人脸位置
# 显示结果图
if answer_pic == True: # 得到结果图
left_top = (dlib.rectangle.left(d), dlib.rectangle.top(d))
right_bottom = (dlib.rectangle.right(d),
dlib.rectangle.bottom(d))
left_top_bottom = (dlib.rectangle.left(d),
dlib.rectangle.bottom(d)) # 作为解释文字的左上坐标
cv2.namedWindow("img", 0)
cv2.rectangle(img, left_top, right_bottom, (0, 255, 0), 2,
cv2.LINE_AA)
# putText参数:照片 / 添加的文字 / 左上角坐标 / 字体 / 字体大小 / 颜色 / 字体粗细
# putText(img, text, org, fontFace, fontScale, color, thickness=None, lineType=None,
# bottomLeftOrigin=None):
cv2.putText(img, cd_sorted[0][0], left_top_bottom,
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 255), 2)
cv2.imshow("img", cv2.cvtColor(img,
cv2.COLOR_RGB2BGR)) # 转成ＢＧＲ显示
# if cd_sorted[0][1] > 0.4:
# answer = "Cannot find a person in the label database!"
# else:
# answer = "The person is:"+cd_sorted[0][0]
# print("\n The person is:", cd_sorted[0][0])
# if answer is not None:
# answer.show_information()
# dlib.hit_enter_to_continue()
if len(answer) == 0:
answer = "No Match"
else:
if len(answer) == 1:
answer = '1 person found: ' + ''.join(answer)
else:
answer = str(len(answer)) + ' persons found: ' + ','.join(answer)
print(answer)
return answer
def show_picture_and_shape(self):
"""Shows the image and the face shape on the screen."""
win = dlib.image_window()
win.set_image(self.img)
win.add_overlay(self.__detection_obj)
win.wait_until_closed()
def __init__(self, act=True):
self.wi = dlib.image_window() if act else None
self.framecount = 0
" http://dlib.net/files/shape_predictor_5_face_landmarks.dat.bz2"\n
" http://dlib.net/files/dlib_face_recognition_resnet_model_v1.dat.bz2")
exit()
predictor_path = sys.argv[1]
face_rec_model_path = sys.argv[2]
faces_folder_path = sys.argv[3]
# Load all the models we need: a detector to find the faces
# a shape predictor to find face landmarks so we can precisely localize the face
# and finally the face recognition model.
detector = dlib.get_frontal_face_detector()
sp = dlib.shape_predictor(predictor_path)
facerec = dlib.face_recognition_model_v1(face_rec_model_path)
win = dlib.image_window()
# Now process all the images
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
win.clear_overlay()
win.set_image(img)
# Ask the detector to find the bounding boxes of each face. The 1 in the
# second argument indicates that we should upsample the image 1 time. This
# will make everything bigger and allow us to detect more faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
import cv2
import numpy as np
import dlib
import pandas
from neural_net_svm import NeuralNetSvm
from sklearn.utils import shuffle
from pathlib import Path
import os
import csv
# to store all the frames with the bounding box over the localized faces
# use HoG to detect faces
find_face = dlib.get_frontal_face_detector()
# display the image with the bounding box
win_show_image = dlib.image_window()
# load pre-trained models from dlib to detect the facial landmarks
shape_predictor = 'E:\Wenger\shape_predictor_68_face_landmarks.dat\shape_predictor_68_face_landmarks.dat'
# create the detector using the above pretrained model
face_pose_detector = dlib.shape_predictor(shape_predictor)
# text file opened to save the landmarks
fp = open("landmarks_curious.txt", "wb")
def localize_face_in_frame(path_in, landmarks=[]): #to store landmarks
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
# landmarks index
"execute this program by running:\n"
" ./face_landmark_detection.py shape_predictor_68_face_landmarks.dat ../examples/faces\n"
"You can download a trained facial shape predictor from:\n"
" http://sourceforge.net/projects/dclib/files/dlib/v18.10/shape_predictor_68_face_landmarks.dat.bz2")
exit()
predictor_path = sys.argv[1] # takes in a shape predictor as first argument
faces_folder_path = sys.argv[2] # takes in the folder containing the .jpg images
detector = dlib.get_frontal_face_detector() # detects faces, detector.length = # of faces
# From API: This object (shape_predictor()) is a tool that takes in an image region containing some object and outputs
# a set of point locations that define the pose of the object.
predictor = dlib.shape_predictor(predictor_path)
win = dlib.image_window() # GUI object
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")): # glob alows for '*.jpg' to work
print("Processing file: {}".format(f))
img = io.imread(f)
win.clear_overlay()
win.set_image(img)
# Ask the detector to find the bounding boxes of each face. The 1 in the
# second argument indicates that we should upsample the image 1 time. This
# will make everything bigger and allow us to detect more faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets))) # prints the amount of faces detected
for k, d in enumerate(dets): #for (k = 0; k < dets.length; k ++) where d = dets[k]
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format( # prints the corner coordinates
def draw_rectangle(image, coords):
img = dlib.image_window()
img.set_image(image)
for i, mtr in enumerate(coords):
img.add_overlay(mtr)
return img
from typing import TypeVar, NewType, Type, Generic#, ClassVar
from typing import Container, Hashable, Iterable, Sized, Callable, Awaitable, AsyncIterable
from typing import Iterator, Sequence, Set, Mapping, MappingView, AsyncIterator#, Coroutine
from typing import Generator, MutableSequence, ByteString, MutableSet, MutableMapping, ItemsView, KeysView, ValuesView
from typing import SupportsInt, SupportsFloat, SupportsAbs, SupportsRound
from typing import Union, Optional, AbstractSet, Reversible
from typing import Any, re, io, AnyStr, Tuple, NamedTuple, List, Dict, DefaultDict
import dlib
from skimage import io
from PIL import Image
import numpy as np
import cv2
image_file = 'capture.png' # type: string
img = io.imread(image_file) # type: np.ndarray
# http://kivantium.hateblo.jp/entry/2015/07/25/184346
# selective search
rects = [] # type: List<(bool, {left:Callable[[],int], top:Callable[[],int], right:Callable[[],int], bottom:Callable[[],int]})>
dlib.find_candidate_object_locations(img, rects, min_size=1000)
win = dlib.image_window() # type: dlib.image_window
win.set_image(img)
for k, d in enumerate(rects):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
win.add_overlay(d)
dlib.hit_enter_to_continue()
arduino = serial.Serial('/dev/ttyACM0', 9600)
#abrir webcam
cap = cv2.VideoCapture(0)
#cargar los archivos svm entrenados
detector1 = dlib.fhog_object_detector("dataset_abierto.svm")
detector2 = dlib.fhog_object_detector("dataset_cerrado.svm")
detector3 = dlib.fhog_object_detector("dataset_derecho.svm")
detector4 = dlib.fhog_object_detector("dataset_izquierdo.svm")
detectors = [detector1, detector2, detector3, detector4]
#se muestran en pantalla las caracteristicas de HOG para cada detector
win_det = dlib.image_window()
win_det.set_image(detectors[0])
win_det2 = dlib.image_window()
win_det2.set_image(detectors[1])
win_det3 = dlib.image_window()
win_det3.set_image(detectors[2])
win_det4 = dlib.image_window()
win_det4.set_image(detectors[3])
win = dlib.image_window()
action = "nada"
font_color = (0, 0, 0)
def DetectFaceInListDlib(frameList,
faceDetector=None,
skipLength=2,
debug=False):
'''
Given a frame list, detect (track) the faces
Returns subimages of faces after normalization and smoothing the enclosing rectangle
'''
if ((faceDetector is None)):
predictorPath = 'coreData/shape_predictor_68_face_landmarks.dat'
faceDetector = dlib.get_frontal_face_detector()
if (debug):
win = dlib.image_window()
win.clear_overlay()
faceList = []
newFrameList = []
rowList = []
colList = []
detsList = []
smoothRowSize = []
smoothColSize = []
winSize = (20 / skipLength)
for i in range(0, frameList.shape[0], skipLength):
frame = frameList[i]
dets = faceDetector(frame, 1)
dets = list(enumerate(dets))
if (len(dets) != 1):
continue
detsList.append(dets)
newFrameList.append(frame)
for k, d in (dets):
rowList.append(np.abs(d.left() - d.right()))
colList.append(np.abs(d.top() - d.bottom()))
for i in range(len(rowList)):
rowAvg = np.mean(
rowList[int(max(0, i - winSize)):int(min(len(rowList), i +
winSize))]) + 6
colAvg = np.mean(
colList[int(max(0, i - winSize)):int(min(len(colList), i +
winSize))]) + 6
smoothRowSize.append(int(round(rowAvg)))
smoothColSize.append(int(round(colAvg)))
for i in range(len(detsList)):
dets = detsList[i]
frame = newFrameList[i]
grayFrame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
rowc = (d.left() + d.right()) / 2
colc = (d.top() + d.bottom()) / 2
rows = smoothRowSize[i]
cols = smoothColSize[i]
# Forcefully make the enclosing box a square
rows = max(rows, cols)
cols = max(rows, cols)
faceImg = grayFrame[
int(max(0, colc -
(cols / 2))):int(min(frame.shape[0], colc + (cols / 2) +
1)),
int(max(0, rowc -
(rows / 2))):int(min(frame.shape[1], rowc + (rows / 2) +
1))]
# Smoothing rectangle sizes (Running average of -50/+50 frames)
faceImg = cv2.equalizeHist(faceImg)
# Illumination (CLAHE normalization) Normalization
if debug:
win.clear_overlay()
grayFrame[int(max(0, colc - (
cols / 2))):int(min(frame.shape[0], colc + (cols / 2) + 1)),
int(max(0, rowc - (rows / 2))
):int(min(frame.shape[1], rowc + (rows / 2) +
1))] = faceImg
win.set_image(grayFrame)
dlib.hit_enter_to_continue()
faceImg = cv2.resize(faceImg, (100, 100))
faceImg = np.array(faceImg)
faceList.append(faceImg)
faceList = np.array(faceList)
print("该视频总共检测出", faceList.shape[0], "个face_in_list(dlib)文件")
return faceList
def train(image_folder, append):
if append == 0:
#code to open createXML for the noob user with all the required params
cmd = 'createXML.exe -c' + image_folder + '/training.xml ' + image_folder
run(cmd, 5)
cmd = 'createXML.exe ' + image_folder + '/training.xml'
run(cmd, 50)
# <NOTE> <IN PROGRESS> include code to write new XML to the old XML and use the latter for the training
elif append == 1:
#code to open createXML for the noob user with all the required params
cmd = 'createXML.exe -c' + image_folder + '/trainingTemp.xml ' + image_folder
run(cmd, 5)
cmd = 'createXML.exe ' + image_folder + '/trainingTemp.xml'
run(cmd, 50)
dlib.hit_enter_to_continue()
# doing all the magic stuff to append the new XML to the old one
xml1 = image_folder + "/training.xml"
xml2 = image_folder + "/trainingTemp.xml"
removeUselessText(xml1)
removeUselessText(xml2)
#combineXML(xml1,xml2)
r = XMLCombiner((xml1, xml2)).combine()
with open(xml1, "r+") as f:
f.write(et.tostring(r.getroot()))
#Convert the XML to better format before saving it for the training as there may be some improper indentation
# setting option in dlib
options = dlib.simple_object_detector_training_options()
# symmetric detector
options.add_left_right_image_flips = True
# SVM C parameter.larger value will lead to overfitting
options.C = 2
# Tell the code how many CPU cores your computer has for the fastest training.
options.num_threads = 4
options.be_verbose = True
training_xml_path = os.path.join(image_folder, "training.xml")
#testing_xml_path = os.path.join(image_folder, "testing.xml")
# saving the detector as detector.svm with input as the xml file after doing the training
dlib.train_simple_object_detector(training_xml_path, "detector.svm",
options)
# Printing the accuracy with training data
print("\nTraining accuracy: {}".format(
dlib.test_simple_object_detector(training_xml_path, "detector.svm")))
# Doing the detection
detector = dlib.simple_object_detector("detector.svm")
# Looking at the HOG filter the machine has learned.
win_det = dlib.image_window()
win_det.set_image(detector)
def grab_boundingbox_face_test(image_directory):
'''
http://dlib.net/face_landmark_detection.py.html where I got this from
Grabs faces and important features to feed into network from the test images
:param training_image_directory: Directory of the images
:return: bounding boxes of every face and 68 features of every face as well as what faces were not grabbed
'''
import dlib
import glob
data_directory = get_data_directory()
predictor_path = data_directory + "/dlib-models/shape_predictor_68_face_landmarks.dat"
faces_folder_path = image_directory
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_path)
win = dlib.image_window()
shape_list = []
bounding_face_list = []
# If it is a 1 in the list then it was excluded and a 0 means included
# Going to be used when grabbing labels and when figuring out what to bound
exclusion_list = []
current_index = 0
for f in glob.glob(os.path.join(faces_folder_path, "*.pgm")):
print("Processing file: {}".format(f))
# TODO: FIX THIS
img = Image.open(f).convert('L')
img = np.array(img, dtype=np.uint8)
print(np.shape(img))
#img = dlib.load_grayscale_image(f)
# img = dlib.load_rgb_image(f)
# Do this now before we try finding a face
exclusion_list.append(1)
win.clear_overlay()
win.set_image(img)
# Ask the detector to find the bounding boxes of each face. The 1 in the
# second argument indicates that we should upsample the image 1 time. This
# will make everything bigger and allow us to detect more faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
# Found a face so make this zero
exclusion_list[current_index] = 0
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
bounding_face_list.append(d)
# Get the landmarks/parts for the face in box d.
shape = predictor(img, d)
print("Part 0: {}, Part 1: {} ...".format(shape.part(0),
shape.part(1)))
shape_list.append(shape)
# Draw the face landmarks on the screen.
win.add_overlay(shape)
current_index += 1
win.add_overlay(dets)
bounding_face_list = np.array(bounding_face_list)
shape_list = np.array(shape_list)
# Need to reshape to be (len, 1) not (len,)
# bounding_face_list = bounding_face_list.reshape(len(bounding_face_list))
# shape_list = shape_list.reshape(len(shape_list))
# TODO: Comment these out before turning in
print(np.shape(bounding_face_list))
print(np.shape(shape_list))
print(exclusion_list)
print(np.shape(bounding_face_list))
return bounding_face_list, shape_list, exclusion_list
def grimace(filename):
import argparse,sys
try:
from FeatureGen import findRatio, generateFeatures
except ImportError:
exit()
try:
import dlib
from skimage import io
import numpy
import cv2
from sklearn.externals import joblib
except ImportError:
exit()
emotions={ 1:"Anger", 2:"Contempt", 3:"Disgust", 4:"Sleep", 5:"Happy", 6:"Sadness", 7:"Cry"}
def Predict_Emotion(filename):
try:
img=io.imread(filename)
cvimg=cv2.imread(filename)
except:
return
win.clear_overlay()
win.set_image(img)
dets=detector(img,1)
if len(dets)==0:
print "Unable to find any face."
return
for k,d in enumerate(dets):
shape=predictor(img,d)
landmarks=[]
for i in range(68):
landmarks.append(shape.part(i).x)
landmarks.append(shape.part(i).y)
landmarks=numpy.array(landmarks)
features=generateFeatures(landmarks)
features= numpy.asarray(features)
pca_features=pca.transform(features)
emo_predicts=classify.predict(pca_features)
global fin
fin = emotions[int(emo_predicts[0])]
print 'Grimace:', fin
#font = cv2.FONT_HERSHEY_SIMPLEX
#cv2.putText(cvimg,emotions[int(emo_predicts[0])],(20,20), font, 1,(0,255,255),2)
win.add_overlay(shape)
cv2.namedWindow("Output")
cv2.imshow("Output",cvimg)
time.sleep(2)
cv2.destroyAllWindows()
if __name__ == "__main__":
landmark_path="shape_predictor_68_face_landmarks.dat"
detector= dlib.get_frontal_face_detector()
try:
predictor= dlib.shape_predictor(landmark_path)
except:
exit()
win=dlib.image_window()
try:
classify=joblib.load("traindata.pkl")
pca=joblib.load("pcadata.pkl")
except:
exit()
Predict_Emotion(filename)
histr = cv2.calcHist([img], [i], None, [256], [0, 256])
plt.plot(histr, color=col)
plt.xlim([0, 256])
plt.show()
def plot_hist(self, hist):
plt.plot(hist)
plt.xlim([0, 256])
plt.show()
if __name__ == '__main__':
fe = Face_Encoding()
win1 = dlib.image_window()
win2 = dlib.image_window()
win3 = dlib.image_window()
win4 = dlib.image_window()
for image_path in paths.list_images(
"D:\Tuts\DataScience\Python\Datasets\FGNET\Age_Test\Old"):
image = dlib.load_rgb_image(image_path)
gray = dlib.load_grayscale_image(image_path)
#image = cv2.resize(image, (300, 300))
#gray = cv2.resize(gray, (300, 300))
descriptor, image = fe._compute_facenet_embedding_dlib(image=image,
draw=True)
def finding_face_landmark(file_name):
# You can download the required pre-trained face detection model here:
# http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
model = "shape_predictor_68_face_landmarks.dat"
face_detector = dlib.get_frontal_face_detector()
shape_predictor = dlib.shape_predictor(model)
image_window = dlib.image_window()
image = io.imread(file_name)
detected_faces = face_detector(image, 1)
print("Found {} faces in the image file {}".format(len(detected_faces),
image))
if (len(detected_faces) != 1):
print(
"On the photo, there are more faces. Please try out with different photo. "
)
return []
image_window.set_image(image)
face = detected_faces[0]
image_window.add_overlay(face)
landmarks = shape_predictor(image, face)
leftEye1 = landmarks.part(42)
rightEye1 = landmarks.part(39)
nose = landmarks.part(30)
noseTip = landmarks.part(27)
mouth = landmarks.part(62)
noseLeft = landmarks.part(31)
noseRight = landmarks.part(35)
right1 = landmarks.part(1)
left1 = landmarks.part(15)
right2 = landmarks.part(4)
left2 = landmarks.part(12)
right3 = landmarks.part(6)
left3 = landmarks.part(10)
leftEye2 = landmarks.part(45)
rightEye2 = landmarks.part(36)
# for j in range(1, 68):
# pos = pose_landmarks.part(j)
# cv2.circle(image, (pos.x, pos.y), 1, (0, 0, 255), -1)
# cv2.circle(image, (leftEye1.x, leftEye1.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (rightEye1.x, rightEye1.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (nose.x, nose.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (noseTip.x, noseTip.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (mouth.x, mouth.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (noseLeft.x, noseLeft.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (noseRight.x, noseRight.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (right1.x, right1.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (right2.x, right2.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (right3.x, right3.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (left1.x, left1.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (left2.x, left2.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (left3.x, left3.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (leftEye2.x, leftEye2.y), 1, (255, 255, 0), -1)
# cv2.circle(image, (rightEye2.x, rightEye2.y), 1, (255, 255, 0), -1)
# cv2.line(image, (leftEye1.x, leftEye1.y), (rightEye1.x, rightEye1.y), (255, 255, 0), 1)
# cv2.line(image, (leftEye1.x, leftEye1.y), (mouth.x, mouth.y), (255, 255, 0), 1)
# cv2.line(image, (rightEye1.x, rightEye1.y), (mouth.x, mouth.y), (255, 255, 0), 1)
# cv2.line(image, (leftEye1.x, leftEye1.y), (nose.x, nose.y), (255, 255, 0), 1)
# cv2.line(image, (rightEye1.x, rightEye1.y), (nose.x, nose.y), (255, 255, 0), 1)
# cv2.line(image, (mouth.x, mouth.y), (nose.x, nose.y), (255, 255, 0), 1)
# cv2.line(image, (noseTip.x, noseTip.y), (nose.x, nose.y), (255, 255, 0), 1)
# cv2.line(image, (noseLeft.x, noseLeft.y), (noseRight.x, noseRight.y), (255, 255, 0), 1)
# cv2.line(image, (left1.x, left1.y), (right1.x, right1.y), (255, 255, 0), 1)
# cv2.line(image, (left2.x, left2.y), (right2.x, right2.y), (255, 255, 0), 1)
# cv2.line(image, (left3.x, left3.y), (right3.x, right3.y), (255, 255, 0), 1)
# cv2.line(image, (left1.x, left1.y), (leftEye2.x, leftEye2.y), (255, 255, 0), 1)
# cv2.line(image, (right1.x, right1.y), (rightEye2.x, rightEye2.y), (255, 255, 0), 1)
# cv2.line(image, (leftEye1.x, leftEye1.y), (leftEye2.x, leftEye2.y), (255, 255, 0), 1)
# cv2.line(image, (rightEye1.x, rightEye1.y), (rightEye2.x, rightEye2.y), (255, 255, 0), 1)
d1 = euclidean_distance(leftEye1, rightEye1) # distance between the eyes
d2 = euclidean_distance(
leftEye1, mouth
) # distance between middle of the left eyes and middle point of mouth
d3 = euclidean_distance(
rightEye1, mouth
) # distance between middle of the right eyes and middle point of mouth
d4 = euclidean_distance(
leftEye1, nose
) # distance between middle of the left eyes and middle point of nose
d5 = euclidean_distance(
rightEye1, nose
) # distance between middle of the rigth eyes and middle point of nose
d6 = euclidean_distance(
mouth, nose
) # distance between middle point of mouth and middle point of nose
d7 = euclidean_distance(
noseTip, nose) # distance of middle point of d1 and middle of nose
d8 = euclidean_distance(noseLeft, noseRight) # width of nose
d9 = euclidean_distance(left1, right1) # width of face
d10 = euclidean_distance(left2, right2) # width of face
d11 = euclidean_distance(left3, right3) # width of face
d12 = euclidean_distance(leftEye1, leftEye2) # width od left eye
d13 = euclidean_distance(rightEye1, rightEye2) # width of right eye
d14 = euclidean_distance(left1, leftEye2)
d15 = euclidean_distance(right1, rightEye2)
features = []
features.append(d1)
features.append(d2)
features.append(d3)
features.append(d4)
features.append(d5)
features.append(d6)
features.append(d7)
features.append(d8)
features.append(d9)
features.append(d10)
features.append(d11)
features.append(d12)
features.append(d13)
features.append(d14)
features.append(d15)
image_window.add_overlay(landmarks)
# cv2.imshow("Output", image)
# cv2.waitKey(0)
dlib.hit_enter_to_continue()
return features
print("") # Print blank line to create gap from previous output
print("Training accuracy: {}".format(
dlib.test_simple_object_detector(training_xml_path, "detector.svm")))
# However, to get an idea if it really worked without overfitting we need to
# run it on images it wasn't trained on. The next line does this. Happily, we
# see that the object detector works perfectly on the testing images.
print("Testing accuracy: {}".format(
dlib.test_simple_object_detector(testing_xml_path, "detector.svm")))
# Now let's use the detector as you would in a normal application. First we
# will load it from disk.
detector = dlib.simple_object_detector("detector_eye.svm")
#detector = dlib.get_frontal_face_detector()
# We can look at the HOG filter we learned. It should look like a face. Neat!
win_det = dlib.image_window()
win_det.set_image(detector)
## Now let's run the detector over the images in the faces folder and display the
## results.
print("Showing detections on the images in the faces folder...")
win = dlib.image_window()
for f in glob.glob(os.path.join(faces_folder, "BioID_0102.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
dets = detector(img)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
def __init__(self):
self.detector = dlib.get_frontal_face_detector()
self.win = dlib.image_window()
def main():
#Defining the video capture object
video_capture = cv2.VideoCapture(1)
thresh = 0.25
frame_check = 15
detect = dlib.get_frontal_face_detector()
predict = dlib.shape_predictor("/home/agopinath1996/git_ws/deepgaze/scripts/shape_predictor_68_face_landmarks.dat")# change to path where landmark points are stored
(lStart, lEnd) = face_utils.FACIAL_LANDMARKS_68_IDXS["left_eye"] # get the left eye index
(rStart, rEnd) = face_utils.FACIAL_LANDMARKS_68_IDXS["right_eye"] # get the right eye index
flag=0
sess = tf.Session()
my_head_pose_estimator = CnnHeadPoseEstimator(sess)
my_head_pose_estimator.load_roll_variables(os.path.realpath("/home/agopinath1996/git_ws/deepgaze/etc/tensorflow/head_pose/roll/cnn_cccdd_30k.tf"))# change to deepgaze directory path
my_head_pose_estimator.load_pitch_variables(os.path.realpath("/home/agopinath1996/git_ws/deepgaze/etc/tensorflow/head_pose/pitch/cnn_cccdd_30k.tf"))
my_head_pose_estimator.load_yaw_variables(os.path.realpath("/home/agopinath1996/git_ws/deepgaze/etc/tensorflow/head_pose/yaw/cnn_cccdd_30k.tf"))
#Start of Eye Gaze Tracking
win = dlib.image_window()
predictor_path = "/home/agopinath1996/git_ws/deepgaze/scripts/shape_predictor_68_face_landmarks.dat"
roi = []
ref_point = 0
index1 = 0
pt_lefteye_corner_x= 0
pt_lefteye_corner_y = 0
pt_pos1 = 0
predictor = dlib.shape_predictor(predictor_path)
pt_x2 =0
pt_y2 = 0
pt_x1 = 0
pt_y1 = 0
pt_actualx = 0
pt_actualy = 0
detector = dlib.get_frontal_face_detector()
flag = 0
flag1 = 0
pt_righteye_corner_x = 0
pt_righteye_corner_y = 0
if(video_capture.isOpened() == False):
print("Error: the resource is busy or unvailable")
else:
print("The video source has been opened correctly...")
#Create the main window and move it
cv2.namedWindow('Video')
cv2.moveWindow('Video', 20, 20)
#Obtaining the CAM dimension
cam_w = int(video_capture.get(3))
cam_h = int(video_capture.get(4))
#Defining the camera matrix.
#To have better result it is necessary to find the focal
# lenght of the camera. fx/fy are the focal lengths (in pixels)
# and cx/cy are the optical centres. These values can be obtained
# roughly by approximation, for example in a 640x480 camera:
# cx = 640/2 = 320
# cy = 480/2 = 240
# fx = fy = cx/tan(60/2 * pi / 180) = 554.26
c_x = cam_w / 2
c_y = cam_h / 2
f_x = c_x / numpy.tan(60/2 * numpy.pi / 180)
f_y = f_x
#Estimated camera matrix values.
camera_matrix = numpy.float32([[f_x, 0.0, c_x],
[0.0, f_y, c_y],
[0.0, 0.0, 1.0] ])
print("Estimated camera matrix: \n" + str(camera_matrix) + "\n")
#These are the camera matrix values estimated on my webcam with
# the calibration code (see: src/calibration):
camera_matrix = numpy.float32([[602.10618226, 0.0, 320.27333589],
[ 0.0, 603.55869786, 229.7537026],
[ 0.0, 0.0, 1.0] ])
#Distortion coefficients
#camera_distortion = numpy.float32([0.0, 0.0, 0.0, 0.0, 0.0])
#Distortion coefficients estimated by calibration
camera_distortion = numpy.float32([ 0.06232237, -0.41559805, 0.00125389, -0.00402566, 0.04879263])
#This matrix contains the 3D points of the
# 11 landmarks we want to find. It has been
# obtained from antrophometric measurement
# on the human head.
landmarks_3D = numpy.float32([P3D_RIGHT_SIDE,
P3D_GONION_RIGHT,
P3D_MENTON,
P3D_GONION_LEFT,
P3D_LEFT_SIDE,
P3D_FRONTAL_BREADTH_RIGHT,
P3D_FRONTAL_BREADTH_LEFT,
P3D_SELLION,
P3D_NOSE,
P3D_SUB_NOSE,
P3D_RIGHT_EYE,
P3D_RIGHT_TEAR,
P3D_LEFT_TEAR,
P3D_LEFT_EYE,
P3D_STOMION])
#Declaring the two classifiers
my_cascade = haarCascade("/home/agopinath1996/git_ws/deepgaze/etc/xml/haarcascade_frontalface_alt.xml", "/home/agopinath1996/git_ws/deepgaze/etc/xml/haarcascade_profileface.xml")
#TODO If missing, example file can be retrieved from http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
my_detector = faceLandmarkDetection('/home/agopinath1996/git_ws/deepgaze/scripts/shape_predictor_68_face_landmarks.dat')
#Error counter definition
no_face_counter = 0
#Variables that identify the face
#position in the main frame.
face_x1 = 0
face_y1 = 0
face_x2 = 0
face_y2 = 0
face_w = 0
face_h = 0
#Variables that identify the ROI
#position in the main frame.
roi_x1 = 0
roi_y1 = 0
roi_x2 = cam_w
roi_y2 = cam_h
roi_w = cam_w
roi_h = cam_h
roi_resize_w = int(cam_w/10)
roi_resize_h = int(cam_h/10)
while(True):
# Capture frame-by-frame
ret, frame = video_capture.read()
ret, frame_eye = video_capture.read()
#print("Estimated [roll, pitch, yaw] ..... [" + str(roll[0,0,0]) + "," + str(pitch[0,0,0]) + "," + str(yaw[0,0,0]) + "]")
gray = cv2.cvtColor(frame[roi_y1:roi_y2, roi_x1:roi_x2], cv2.COLOR_BGR2GRAY)
img = cv2.cvtColor(frame_eye, cv2.COLOR_RGB2BGR) #for eye gaze detection
drowsyframe = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
subjects = detect(drowsyframe, 0)
for subject in subjects:
shape = predict(frame,subject)
shape = face_utils.shape_to_np(shape)
leftEye = shape[lStart:lEnd]
rightEye = shape[rStart:rEnd]
leftEAR = eye_aspect_ratio(leftEye)
rightEAR = eye_aspect_ratio(rightEye)
ear = leftEAR + rightEAR / 2.0
leftEyeHull = cv2.convexHull(leftEye)
rightEyeHull = cv2.convexHull(rightEye)
cv2.drawContours(frame, [leftEyeHull], -1, (0,255,0),1)
cv2.drawContours(frame, [rightEyeHull], -1, (0,255,0), 1)
if ear<thresh:
flag+= 1
print(flag)
if flag >= frame_check:
cv2.putText(frame, "WAKEUPPPP", (10,30), cv2.FONT_HERSHEY_PLAIN, 1.6, (10,10,255), 2)
cv2.putText(frame, "WAKEUPPPP", (10, 325), cv2.FONT_HERSHEY_PLAIN, 1.6, (10,10,255),2)
else:
flag=0
#Eye Gaze Detetction
dets = detect(img, 0)
check = 5
shapes_eye = []
for k,d in enumerate(dets):
#print("dets{}".format(d))
#print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(k, d.left(), d.top(), d.right(), d.bottom()))
shape_eye = predict(img, d)
for index, pt in enumerate(shape_eye.parts()):
#print('Part {}: {}'.format(index, pt))
pt_pos = (pt.x, pt.y)
cv2.circle(img, pt_pos, 1, (0,225, 0), 2)
if index == 29:
pt_x2 = int(pt.x)
pt_y2 = int(pt.y)
if index == 18:
pt_x1 = int(pt.x)
pt_y1 = int(pt.y)
if index == 37:
pt_righteye_corner_x = pt.x
pt_righteye_corner_y = pt.y
if index == 40:
pt_lefteye_corner_x = pt.x
pt_lefteye_corner_y = pt.y
roi = frame_eye[pt_y1:pt_y2,pt_x1:pt_x2]
roi_gray = cv2.cvtColor(roi,cv2.COLOR_RGB2GRAY)
_, threshold = cv2.threshold(roi_gray, 30, 255, cv2.THRESH_BINARY_INV)
try:
M = cv2.moments(threshold)
#print(M)
cX = int(M["m10"]/M["m00"])
cY = int(M["m01"]/M["m00"])
#print(cX,cY)
pt_actualx = pt_x1+cX
pt_actualy = pt_y1+cY
#print(pt_actualx,pt_actualy)
diff_right = pt_actualx-pt_righteye_corner_x
diff_left = pt_lefteye_corner_x - pt_actualx
print(diff_right,diff_left)
#print(cX,cY)
if diff_right < 3:
cv2.putText(frame,'Look straight!',(10,60),cv2.FONT_HERSHEY_SIMPLEX,0.5,(10,10,255),2)
if diff_left <3:
cv2.putText(frame,'Look straight!',(10,60),cv2.FONT_HERSHEY_SIMPLEX,0.5,(10,10,255),2)
except:
pass
cv2.circle(frame,(pt_actualx,pt_actualy), 2,(255,0,255),-1)
#print(pt_actualx,pt_actualy)
#print(pt_x1,pt_x2,pt_y1,pt_y2)
#print(roi.shape_eye)
#print(img.shape_eye)
try:
cv2.imshow("threshold", threshold)
cv2.waitKey(1)
except:
pass
win.clear_overlay()
win.set_image(img)
if len(shapes_eye)!= 0 :
for i in range(len(shapes_eye)):
win.add_overlay(shapes_eye[i])
#Looking for faces with cascade
#The classifier moves over the ROI
#starting from a minimum dimension and augmentig
#slightly based on the scale factor parameter.
#The scale factor for the frontal face is 1.10 (10%)
#Scale factor: 1.15=15%,1.25=25% ...ecc
#Higher scale factors means faster classification
#but lower accuracy.
#
#Return code: 1=Frontal, 2=FrontRotLeft,
# 3=FrontRotRight, 4=ProfileLeft, 5=ProfileRight.
my_cascade.findFace(gray, True, True, True, True, 1.10, 1.10, 1.15, 1.15, 40, 40, rotationAngleCCW=30, rotationAngleCW=-30, lastFaceType=my_cascade.face_type)
#print(returnvalue)
#Accumulate error values in a counter
if(my_cascade.face_type == 0):
no_face_counter += 1
#If any face is found for a certain
#number of cycles, then the ROI is reset
if(no_face_counter == 50):
no_face_counter = 0
roi_x1 = 0
roi_y1 = 0
roi_x2 = cam_w
roi_y2 = cam_h
roi_w = cam_w
roi_h = cam_h
#Checking wich kind of face it is returned
if(my_cascade.face_type > 0):
#Face found, reset the error counter
no_face_counter = 0
#Because the dlib landmark detector wants a precise
#boundary box of the face, it is necessary to resize
#the box returned by the OpenCV haar detector.
#Adjusting the frame for profile left
if(my_cascade.face_type == 4):
face_margin_x1 = 20 - 10 #resize_rate + shift_rate
face_margin_y1 = 20 + 5 #resize_rate + shift_rate
face_margin_x2 = -20 - 10 #resize_rate + shift_rate
face_margin_y2 = -20 + 5 #resize_rate + shift_rate
face_margin_h = -0.7 #resize_factor
face_margin_w = -0.7 #resize_factor
#Adjusting the frame for profile right
elif(my_cascade.face_type == 5):
face_margin_x1 = 20 + 10
face_margin_y1 = 20 + 5
face_margin_x2 = -20 + 10
face_margin_y2 = -20 + 5
face_margin_h = -0.7
face_margin_w = -0.7
#No adjustments
else:
face_margin_x1 = 0
face_margin_y1 = 0
face_margin_x2 = 0
face_margin_y2 = 0
face_margin_h = 0
face_margin_w = 0
#Updating the face position
face_x1 = my_cascade.face_x + roi_x1 + face_margin_x1
face_y1 = my_cascade.face_y + roi_y1 + face_margin_y1
face_x2 = my_cascade.face_x + my_cascade.face_w + roi_x1 + face_margin_x2
face_y2 = my_cascade.face_y + my_cascade.face_h + roi_y1 + face_margin_y2
face_w = my_cascade.face_w + int(my_cascade.face_w * face_margin_w)
face_h = my_cascade.face_h + int(my_cascade.face_h * face_margin_h)
crop_img = frame[face_y1:face_y2, face_x1:face_x2]
cv2.imshow("cropped", crop_img)
roll = my_head_pose_estimator.return_roll(crop_img)
pitch = my_head_pose_estimator.return_pitch(crop_img)
yaw = my_head_pose_estimator.return_yaw(crop_img)
#print("Estimated [roll, pitch, yaw] ..... [" + str(roll[0,0,0]) + "," + str(pitch[0,0,0]) + "," + str(yaw[0,0,0]) + "]")
if yaw > 30:
cv2.putText(frame, "You are facing right!", (10,30), cv2.FONT_HERSHEY_PLAIN, 1.6, (10,10,255), 2)
if yaw < -30:
cv2.putText(frame, "You are facing left!", (10,30), cv2.FONT_HERSHEY_PLAIN, 1.6, (10,10,255), 2)
#Updating the ROI position
roi_x1 = face_x1 - roi_resize_w
if (roi_x1 < 0): roi_x1 = 0
roi_y1 = face_y1 - roi_resize_h
if(roi_y1 < 0): roi_y1 = 0
roi_w = face_w + roi_resize_w + roi_resize_w
if(roi_w > cam_w): roi_w = cam_w
roi_h = face_h + roi_resize_h + roi_resize_h
if(roi_h > cam_h): roi_h = cam_h
roi_x2 = face_x2 + roi_resize_w
if (roi_x2 > cam_w): roi_x2 = cam_w
roi_y2 = face_y2 + roi_resize_h
if(roi_y2 > cam_h): roi_y2 = cam_h
#Debugging printing utilities
if(DEBUG == True):
#print("FACE: ", face_x1, face_y1, face_x2, face_y2, face_w, face_h)
#print("ROI: ", roi_x1, roi_y1, roi_x2, roi_y2, roi_w, roi_h)
#Drawing a green rectangle
# (and text) around the face.
text_x1 = face_x1
text_y1 = face_y1 - 3
if(text_y1 < 0): text_y1 = 0
cv2.putText(frame, "FACE", (text_x1,text_y1), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 1);
cv2.rectangle(frame,
(face_x1, face_y1),
(face_x2, face_y2),
(0, 255, 0),
2)
#In case of a frontal/rotated face it
# is called the landamark detector
if(my_cascade.face_type > 0):
landmarks_2D = my_detector.returnLandmarks(frame, face_x1, face_y1, face_x2, face_y2, points_to_return=TRACKED_POINTS)
if(DEBUG == True):
#cv2.drawKeypoints(frame, landmarks_2D)
for point in landmarks_2D:
cv2.circle(frame,( point[0], point[1] ), 2, (0,0,255), -1)
#Applying the PnP solver to find the 3D pose
# of the head from the 2D position of the
# landmarks.
#retval - bool
#rvec - Output rotation vector that, together with tvec, brings
# points from the model coordinate system to the camera coordinate system.
#tvec - Output translation vector.
retval, rvec, tvec = cv2.solvePnP(landmarks_3D,
landmarks_2D,
camera_matrix, camera_distortion)
#Now we project the 3D points into the image plane
#Creating a 3-axis to be used as reference in the image.
axis = numpy.float32([[50,0,0],
[0,50,0],
[0,0,50]])
imgpts, jac = cv2.projectPoints(axis, rvec, tvec, camera_matrix, camera_distortion)
#Drawing the three axis on the image frame.
#The opencv colors are defined as BGR colors such as:
# (a, b, c) >> Blue = a, Green = b and Red = c
#Our axis/color convention is X=R, Y=G, Z=B
sellion_xy = (landmarks_2D[7][0], landmarks_2D[7][1])
cv2.line(frame, sellion_xy, tuple(imgpts[1].ravel()), (0,255,0), 3) #GREEN
cv2.line(frame, sellion_xy, tuple(imgpts[2].ravel()), (255,0,0), 3) #BLUE
cv2.line(frame, sellion_xy, tuple(imgpts[0].ravel()), (0,0,255), 3) #RED
#Drawing a yellow rectangle
# (and text) around the ROI.
if(DEBUG == True):
text_x1 = roi_x1
text_y1 = roi_y1 - 3
if(text_y1 < 0): text_y1 = 0
cv2.putText(frame, "ROI", (text_x1,text_y1), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,255), 1);
cv2.rectangle(frame,
(roi_x1, roi_y1),
(roi_x2, roi_y2),
(0, 255, 255),
2)
#Showing the frame and waiting
# for the exit command
cv2.imshow('Video', frame)
if cv2.waitKey(1) & 0xFF == ord('q'): break
#Release the camera
video_capture.release()
print("Bye...")
def display_image(img, shape):
win = dlib.image_window()
win.clear_overlay()
win.set_image(img)
if shape:
win.add_overlay(shape)
#open csv 寫入檔
csvname = ImageName + '_record.csv'
csvfile = open(csvname, 'w', newline='')
#設定字典
dic = ['index', 'coordinate_x', 'coordinate_y']
writer = csv.DictWriter(csvfile, fieldnames=dic)
writer.writeheader()
# 使用特徵提取器get_frontal_face_detector
detector = dlib.get_frontal_face_detector()
# dlib的68點模型，使用作者訓練好的特徵預測器
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
# 圖片所在路徑
img = io.imread(ImageName + filetype)
# 生成dlib的圖像窗口
win = dlib.image_window() #能夠在螢幕上顯示圖像的GUI窗口
win.clear_overlay() #從image_window中刪除所有疊加層
win.set_image(img) #重載功能
# 特徵提取器的實例化
dets = detector(img, 1)
print("人臉數：", len(dets))
for k, d in enumerate(dets):
print("第", k + 1, "個人臉d的座標：", "left:", d.left(), "right:", d.right(),
"top:", d.top(), "bottom:", d.bottom())
width = d.right() - d.left()
heigth = d.bottom() - d.top()
print('人臉面積為：', (width * heigth))
# 利用預測器預測
def detect(path):
if len(dataset_matrix) != 0:
del dataset_matrix[:]
win = dlib.image_window()
frame = io.imread(path)
#video_capture = cv2.VideoCapture(0) # for webcam
detector = dlib.get_frontal_face_detector() # Face detector
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
#ret, frame = video_capture.read() #for webcam
win.clear_overlay()
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
clahe_image = clahe.apply(gray)
detections = detector(clahe_image, 1) # Detect the faces in the image
print "Number of faces: {}".format(len(detections))
if len(detections) == 0:
return None, None
#lists for x and y coordinates
xlist = [[] for i in range(len(detections))]
ylist = [[] for i in range(len(detections))]
#lists for mean of x and y coordinates --> gets center of the face
xmean = [[] for i in range(len(detections))]
ymean = [[] for i in range(len(detections))]
#matrixes of xlist and ylist for each face
matrix = [[] for i in range(len(detections))]
#distances of each landmark from the center of the face
xcenter = [[] for i in range(len(detections))]
ycenter = [[] for i in range(len(detections))]
#relative coordinates in range [0,1]
xnorm = [[] for i in range(len(detections))]
ynorm = [[] for i in range(len(detections))]
#lists for relative coordinates of means
xmean_norm = [[] for i in range(len(detections))]
ymean_norm = [[] for i in range(len(detections))]
# lists for relative distances from the center of the face
xcenter_norm = [[] for i in range(len(detections))]
ycenter_norm = [[] for i in range(len(detections))]
#matrixes of final results (contains lists: xcenter, ycenter, eucl_distances, angles)
final = [[] for i in range(len(detections))]
for k, d in enumerate(detections): # For each detected face
shape = predictor(clahe_image, d)
for i in range(0, 68): # 68 landmarks
xlist[k].append(shape.part(i).x)
ylist[k].append(shape.part(i).y)
cv2.circle(frame, (shape.part(i).x, shape.part(i).y),
1, (255, 0, 0),
thickness=2) # draw red dots
xmean[k] = numpy.mean(xlist[k]) #mean of x coordinates
ymean[k] = numpy.mean(ylist[k]) #mean of y coordinates
matrix[k] = numpy.column_stack((xlist[k], ylist[k]))
for x, y in matrix[k]:
cv2.line(frame, (x, y), (int(xmean[k]), int(ymean[k])),
(0, 255, 0),
thickness=1) #draw lines
cv2.circle(frame, (int(xmean[k]), int(ymean[k])),
1, (0, 0, 255),
thickness=2) # draw center dot
xcenter[k] = ([x - xmean[k] for x in xlist[k]]) #distances from the
ycenter[k] = ([y - ymean[k] for y in ylist[k]]) #center of the face
win.set_image(frame)
win.add_overlay(detections)
#print min(xlist[k]), max(xlist[k]), min(ylist[k]), max(ylist[k])
for i in xlist[k]:
xnorm[k].append(
float((i - min(xlist[k]))) / float(
(max(xlist[k]) - min(xlist[k]))))
for i in ylist[k]:
ynorm[k].append(
float((i - min(ylist[k]))) / float(
(max(ylist[k]) - min(ylist[k]))))
#print xnorm[k]
#print ynorm[k]
#print xmean
#print ymean
xmean_norm[k] = float((xmean[k] - min(xlist[k]))) / float(
(max(xlist[k]) - min(xlist[k])))
ymean_norm[k] = float((ymean[k] - min(ylist[k]))) / float(
(max(ylist[k]) - min(ylist[k])))
print "Relative means"
print xmean_norm
print ymean_norm
print "Relative distances"
xcenter_norm[k] = ([x - xmean_norm[k]
for x in xnorm[k]]) # relative distances from the
ycenter_norm[k] = ([y - ymean_norm[k]
for y in ynorm[k]]) # center of the face
print xcenter_norm
print ycenter_norm
for i in range(len(xcenter_norm[k])):
final[k].append(xcenter_norm[k][i])
final[k].append(ycenter_norm[k][i])
#for i in range(len(xcenter_norm[k])):
#h = numpy.hypot(abs(xcenter_norm[k][i]), abs(ycenter_norm[k][i]))
#print h
#final[k].append(h)
emotion = write_emotion(path)
#if emotion != -1:
print EMOTIONS[emotion]
final[k].append(emotion)
#else:
#print ValueError
print final[k]
dataset_matrix.append(final[k])
#write_to_csv(final[k])
print "from detector: {}".format(dataset_matrix)
print final
return frame, dataset_matrix
def train(train, model, test, flips, C, threads, verbose):
# Now let's do the training. The train_simple_object_detector() function has a
# bunch of options, all of which come with reasonable default values. The next
# few lines goes over some of these options.
options = dlib.simple_object_detector_training_options()
#**********************REMOVE!!!!!!!!!!!!!!!!!******************
#Since faces are left/right symmetric we can tell the trainer to train a
# symmetric detector. This helps it get the most value out of the training
# data.
options.add_left_right_image_flips = flips
# The trainer is a kind of support vector machine and therefore has the usual
# SVM C parameter. In general, a bigger C encourages it to fit the training
# data better but might lead to overfitting. You must find the best C value
# empirically by checking how well the trained detector works on a test set of
# images you haven't trained on. Don't just leave the value set at 5. Try a
# few different C values and see what works best for your data.
options.C = C
# Tell the code how many CPU cores your computer has for the fastest training.
options.num_threads = threads
options.be_verbose = verbose
# You just need to put your images into a list. TrainingImages takes a folder
#and extracts the images and bounding boxes for each image
trainingSet = TrainingImages(train)
images = trainingSet.images
# Then for each image you make a list of rectangles which give the pixel
# locations of the edges of the boxes.
# And then you aggregate those lists of boxes into one big list and then call
# train_simple_object_detector().
boxes = trainingSet.boxes
if not images or not boxes:
print "Training set is empty. Check input directory for images and proper \
formating of bounding box files!"
sys.exit(2)
testImages = images
testBoxes = boxes
if test != train:
testSet = TrainingImages(test)
testImages = testSet.images
testBoxes = testSet.boxes
count = 1
width = 0
height = 0
##Calculating boxes, aspect ratios etc. ***May need to adjust logic
##to accommodate ambiguity.
##Also saving new masked images to disk to verify the correct
##annotations are being detected.
if not os.path.exists(train + "/masked"):
os.makedirs(train + "/masked")
aspRatios = []
flatARs = []
dictARs = {}
for j, i in enumerate(boxes):
curImageName = trainingSet.imageNames[j]
print "Image: ", curImageName
newImage = images[j].copy()
aRs = []
for box in i:
cv2.rectangle(newImage, (box.top(), box.left()),
(box.bottom(), box.right()), (255, 255, 255),
thickness=-3)
width += box.width()
height += box.height()
ar = float(box.width()) / float(box.height())
aRs.append(ar)
dictARs[ar] = box
flatARs.append(ar)
count += 1
print "Box: ", box, "\t Area: ", box.area(), "\tAR: ", aRs
aspRatios.append(aRs)
print "\nAspect Ratios: ", aspRatios, "\nDictionary: ", dictARs
baseName = curImageName.split("/")[-1]
newImageName = train + "/masked/" + (baseName).replace(
".jpg", "-boxes.jpg")
print "\nSaving: ", newImageName, "\n"
cv2.imwrite(newImageName, newImage)
##Calculating the mean and standard deviation (May not need mean
##not currently using it...)
aRMean = np.mean(flatARs, 0)
aRStd = np.std(flatARs, 0)
print "Aspect Ratio Mean: ", aRMean, " Std: ", aRStd
target_size = float(width / count) * float(height / count)
#Update the sliding window size based on input data
width, height = PlateDetector.bestWindow(boxes,
target_size=target_size)
targetSize = int(width * height)
targetAr = float(width) / height
options.detection_window_size = targetSize
print "New Width: ", width, "\tNew Height", height, "!!!"
print "Target size: ", targetSize, " Target AR: ", targetAr
##Deleting boxes with aspect ratios that are above or below the target
##aspect ratio plus one standard deviation. bestWindow estimates a target
##aspect ratio close to (but not exactly) the mean. This logic was borrowed
##from a dlib C++ HOG training example. Not sure why they didn't just use the
##mean, but this seems to work fine.
for i, imgArs in enumerate(aspRatios):
for boxArs in imgArs:
if (boxArs > (targetAr + aRStd)) or (boxArs <
(targetAr - aRStd)):
print "Deleting box ", dictARs[boxArs]
boxes[i].remove(dictARs[boxArs])
print "New boxes: ", boxes
#Train
detector = dlib.train_simple_object_detector(images, boxes, options)
# We could save this detector to disk by uncommenting the following.
detector.save(model)
# Now let's look at its HOG filter!
# win_det.set_image(detector)
# dlib.hit_enter_to_continue()
win_det = dlib.image_window()
win_det.set_image(detector)
# Note that you don't have to use the XML based input to
# test_simple_object_detector(). If you have already loaded your training
# images and bounding boxes for the objects then you can call it as shown
# below.
print("\nTraining accuracy: {}".format(
dlib.test_simple_object_detector(testImages, testBoxes, detector)))
users[face_descriptor] = name
p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16,
channels=1,
rate=8000,
input=True,
frames_per_buffer=1024)
stream.start_stream()
in_speech_bf = False
decoder.start_utt()
shape = None
cap = cv2.VideoCapture(-1)
win1 = dlib.image_window()
while True:
voice_command = ""
ret, frame = cap.read()
buf = stream.read(1024)
if buf:
decoder.process_raw(buf, False, False)
if decoder.get_in_speech() != in_speech_bf:
in_speech_bf = decoder.get_in_speech()
if not in_speech_bf:
decoder.end_utt()
try:
voice_command = decoder.hyp().hypstr
except:
voice_command = ""
def DetectFaceInListDlib(frameList, faceDetector = None, skipLength = 2, debug = False): ''' Given a frame list, detect (track) the faces Returns subimages of faces after normalization and smoothing the enclosing rectangle ''' if ((faceDetector is None)): predictorPath = 'coreData/shape_predictor_68_face_landmarks.dat' faceDetector = dlib.get_frontal_face_detector() if (debug): win = dlib.image_window() win.clear_overlay() faceList = [] newFrameList = [] rowList = [] colList = [] detsList = [] smoothRowSize = [] smoothColSize = [] winSize = (20/skipLength) for i in range(0, frameList.shape[0], skipLength): frame = frameList[i] dets = faceDetector(frame, 1) dets = list(enumerate(dets)) if (len(dets) != 1): continue detsList.append(dets) newFrameList.append(frame) for k, d in (dets): rowList.append(np.abs(d.left() - d.right())) colList.append(np.abs(d.top() - d.bottom())) for i in range(len(rowList)): rowAvg = np.mean(rowList[max(0,i-winSize):min(len(rowList),i+winSize)]) + 6 colAvg = np.mean(colList[max(0,i-winSize):min(len(colList),i+winSize)]) + 6 smoothRowSize.append(int(round(rowAvg))) smoothColSize.append(int(round(colAvg))) for i in range(len(detsList)): dets = detsList[i] frame = newFrameList[i] grayFrame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) rowc = (d.left() + d.right())/2 colc = (d.top() + d.bottom())/2 rows = smoothRowSize[i] cols = smoothColSize[i] # Forcefully make the enclosing box a square rows = max(rows, cols) cols = max(rows, cols) faceImg = grayFrame[max(0,colc-(cols/2)):min(frame.shape[0],colc+(cols/2)+1), max(0,rowc-(rows/2)):min(frame.shape[1],rowc+(rows/2)+1)] # Smoothing rectangle sizes (Running average of -50/+50 frames) faceImg = cv2.equalizeHist(faceImg) # Illumination (CLAHE normalization) Normalization if debug: win.clear_overlay() grayFrame[max(0,colc-(cols/2)):min(frame.shape[0],colc+(cols/2)+1), max(0,rowc-(rows/2)):min(frame.shape[1],rowc+(rows/2)+1)] = faceImg win.set_image(grayFrame) dlib.hit_enter_to_continue() faceImg = cv2.resize(faceImg, (100, 100)) faceImg = np.array(faceImg) faceList.append(faceImg) faceList = np.array(faceList) return faceList
def draw_face_landmarks(self,
image,
dets=None,
shapes=[],
return_drawn_landmarks=False,
draw_type="line"):
print("\nDrawing face landmarks..\n")
if not return_drawn_landmarks:
win = dlib.image_window()
win.set_image(image)
if self.sp == self.shape_68_face_landmarks:
face_landmarks_list = face_utils.FACIAL_LANDMARKS_68_IDXS
else:
face_landmarks_list = face_utils.FACIAL_LANDMARKS_5_IDXS
if image is None:
print("Please provide an image")
exit()
if dets is None:
dets = self.detect_face(image=image)
if len(shapes) == 0:
shapes = self.detect_face_landmarks(image=image, dets=dets)
# DOnly raw landmarks and display in dlib window if we are not returning the image
if not return_drawn_landmarks:
for shape in shapes:
win.add_overlay(shape, dlib.rgb_pixel(0, 255, 0))
# Draw landmarks over the image using opencv line or circle to return the drawn image
if return_drawn_landmarks:
for shape in shapes:
shape = face_utils.shape_to_np(shape)
# Loop over the face parts individually
for (name, (i, j)) in face_landmarks_list.items():
# Loop over the subset of facial landmarks, drawing the
# specific face part
px = None
py = None
for (x, y) in shape[i:j]:
if draw_type == "line":
if px is None and py is None:
px, py = x, y
cv2.line(image, (px, py), (x, y), (0, 255, 0), 2)
px, py = x, y
else:
cv2.circle(image, (x, y), 1, (0, 0, 255), -1)
return image
else:
dlib.hit_enter_to_continue()
return image
# However, to get an idea if it really worked without overfitting we need to
# run it on images it wasn't trained on. The next line does this. Happily, we
# see that the object detector works perfectly on the testing images.
print("Testing accuracy: {}".format(
dlib.test_simple_object_detector(testing_xml_path, "detector.svm")))
# Now let's use the detector as you would in a normal application. First we
# will load it from disk.
detector = dlib.simple_object_detector("detector.svm")
# We can look at the HOG filter we learned. It should look like a face. Neat!
win_det = dlib.image_window()
win_det.set_image(detector)
# Now let's run the detector over the images in the faces folder and display the
# results.
print("Showing detections on the images in the faces folder...")
win = dlib.image_window()
for f in glob.glob(os.path.join(faces_folder, "*.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
dets = detector(img)
print("Number of faces detected: {}".format(len(dets)))
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
def get_triangulation(im,
gray_image,
a=50,
b=55,
c=0.15,
show=False,
randomize=False):
'''Returns triangulations'''
# Using canny edge detection.
#
# Reference: http://docs.opencv.org/3.1.0/da/d22/tutorial_py_canny.html
# First argument: Input image
# Second argument: minVal (argument 'a')
# Third argument: maxVal (argument 'b')
#
# 'minVal' and 'maxVal' are used in the Hysterisis Thresholding step.
# Any edges with intensity gradient more than maxVal are sure to be edges
# and those below minVal are sure to be non-edges, so discarded. Those who
# lie between these two thresholds are classified edges or non-edges based
# on their connectivity.
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(predictor_path)
edges = cv2.Canny(gray_image, a, b)
if show:
cv2.imshow('Canny', edges)
cv2.waitKey(0)
cv2.destroyAllWindows()
win = dlib.image_window()
# Set number of points for low-poly edge vertices
num_points = int(np.where(edges)[0].size * c)
# Return the indices of the elements that are non-zero.
# 'nonzero' returns a tuple of arrays, one for each dimension of a,
# containing the indices of the non-zero elements in that dimension.
# So, r consists of row indices of non-zero elements, and c column indices.
r, c = np.nonzero(edges)
# r.shape, here, returns the count of all points that belong to an edge.
# So 'np.zeros(r.shape)' an array of this size, with all zeros.
# 'rnd' is thus an array of this size, with all values as 'False'.
rnd = np.zeros(r.shape) == 1
# Mark indices from beginning to 'num_points - 1' as True.
rnd[:num_points] = True
# Shuffle
np.random.shuffle(rnd)
# Randomly select 'num_points' of points from the set of all edge vertices.
r = r[rnd]
c = c[rnd]
# Number of rows and columns in image
sz = im.shape
r_max = sz[0]
c_max = sz[1]
# Co-ordinates of all randomly chosen points
pts = np.vstack([r, c]).T
if randomize:
rand_offset = 50
rand_dirs = [(0, rand_offset), (-rand_offset, 0), (0, -rand_offset),
(rand_offset, 0)]
rnd_count = 0
for point in pts:
if random.random() < 0.3:
rnd_count += 1
rand_dir = random.randint(0, 3)
point[0] += rand_dirs[rand_dir][0]
point[1] += rand_dirs[rand_dir][1]
# Append (0,0) to the vertical stack
pts = np.vstack([pts, [0, 0]])
# Append (0,c_max) to the vertical stack
pts = np.vstack([pts, [0, c_max]])
# Append (r_max,0) to the vertical stack
pts = np.vstack([pts, [r_max, 0]])
# Append (r_max,c_max) to the vertical stack
pts = np.vstack([pts, [r_max, c_max]])
# Append some random points to fill empty spaces
pts = np.vstack([pts, np.random.randint(0, 750, size=(100, 2))])
# print(len(pts))
# pts = my_reduce(pts, 5)
# print(len(pts))
dets = detector(im, 1)
# print("Number of faces detected: {}".format(len(dets)))
if show:
win.clear_overlay()
win.set_image(im)
for k, d in enumerate(dets):
shape = predictor(im, d)
for i in range(shape.num_parts):
pts = np.vstack([pts, [shape.part(i).x, shape.part(i).y]])
if show:
win.add_overlay(shape)
if show:
win.add_overlay(dets)
dlib.hit_enter_to_continue()
# Construct Delaunay Triangulation from these set of points.
# Reference: https://en.wikipedia.org/wiki/Delaunay_triangulation
tris = Delaunay(pts, incremental=True)
# tris_vertices = pts[tris.simplices]
# for tri in range(tris_vertices.shape[0]):
# x_coords = []
# y_coords = []
# print(tris_vertices[tri])
# for coord in range(tris_vertices.shape[1]):
# x_coords.append(tris_vertices[tri][coord][0])
# y_coords.append(tris_vertices[tri][coord][1])
# divideHighVariance(tris, im)
tris.close()
# exit(0)
# Return triangulation
return tris
# running the command:
# sudo apt-get install libboost-python-dev cmake
#
# Also note that this example requires scikit-image which can be installed
# via the command:
# pip install scikit-image
# Or downloaded from http://scikit-image.org/download.html.
import sys
import dlib
from skimage import io
detector = dlib.get_frontal_face_detector()
win = dlib.image_window()
for f in sys.argv[1:]:
print("Processing file: {}".format(f))
img = io.imread(f)
# The 1 in the second argument indicates that we should upsample the image
# 1 time. This will make everything bigger and allow us to detect more
# faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
for i, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
i, d.left(), d.top(), d.right(), d.bottom()))
win.clear_overlay()
win.set_image(img)
40 E0 D0 which is way lot shorter
How do we change the tone of colors?
We just pass the RGB data of the image and pass it to a mapping
function which changes the tone of the color as described in that function
'''
file_name = "pic4.jpg"
# Create a HOG face detector using the built-in dlib class
face_detector = dlib.get_frontal_face_detector()
#face_detector= dlib.cnn_face_detection_model_v1("mmod_human_face_detector.dat")
# for side face detection
side_face_cascade = cv2.CascadeClassifier('haarcascade_profileface.xml')
win = dlib.image_window() # A window used to display the image
# Load the image into an array
image = io.imread("pic4.jpg")
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# Run the HOG face detector on the image data.
# The result will be the bounding boxes of the faces in our image.
detected_faces = face_detector(image, 1)
side_faces = side_face_cascade.detectMultiScale(gray, 1.1, 5)
print("I found {} front faces in the file {}".format(len(detected_faces),
file_name))
print("I found {}side faces in the file {}".format(len(side_faces),
file_name))