def DrawLandmarks(self, image):
'''
@parms: grayscale image
Predict the lanmarks on image and return the image.
'''
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# detect faces in the grayscale image
rects = self.detector(image, 1)
# loop over the face detections
for (i, rect) in enumerate(rects):
# determine the facial landmarks for the face region, then
# convert the facial landmark (x, y)-coordinates to a NumPy
# array
shape = self.predictor(gray, rect)
shape = utils.shape_to_np(shape)
# convert dlib's rectangle to a OpenCV-style bounding box
# [i.e., (x, y, w, h)], then draw the face bounding box
(x, y, w, h) = utils.rect_to_bb(rect)
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
# show 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)-coordinates for the facial landmarks
# and draw them on the image
for (x, y) in shape:
cv2.circle(image, (x, y), 1, (0, 0, 255), -1)
return image
def lmDetect(self, img, bbox=None):
''' 检测人脸关键点，假定单张图片单人脸
param
img: str or nparray
'''
if type(img) not in [str, np.array]:
raise NotImplementedError('type(img) not supported: {}'.format(
type(img)))
if isinstance(img, str):
try:
img = cv2.imread(img)
rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
except:
print('ERROR read img @lmDetect', img)
if 'dt' in self.detectMode:
bbox = self.faceDetect(img)
if not bbox:
bbox = [0, 0, img.shape[1],
img.shape[0]] # left, top, right, bottom
landmarks = self.predictor(rgb, box=bbox)
if landmarks is None: return None
landmarks = shape_to_np(landmarks)
# import pdb
# pdb.set_trace()
return landmarks
def get_landmarks(detector, predictor, rgb):
# first get bounding box (dlib.rectangle class) of face.
boxes = detector(rgb, 1)
for box in boxes:
landmarks = shape_to_np(predictor(rgb, box=box))
break
else:
return None
return landmarks.astype(np.int32)
def dlibdect(img):
# cv2.imshow('image',img)
# cv2.waitKey(0)
# cv2.destroyAllWindows()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
rects = detector(gray, 1)
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
print(shape.shape)
return shape
def generator_demo_example_lips(img_path):
print(img_path)
name = img_path.split('/')[-1]
landmark_path = os.path.join('../image/', name.replace('jpg', 'npy'))
region_path = os.path.join('../image/',
name.replace('.jpg', '_region.jpg'))
roi, landmark = crop_image(img_path)
if np.sum(landmark[37:39, 1] - landmark[40:42, 1]) < -9:
# pts2 = np.float32(np.array([template[36],template[45],template[30]]))
template = np.load('../basics/base_68.npy')
else:
template = np.load('../basics/base_68_close.npy')
# pts2 = np.float32(np.vstack((template[27:36,:], template[39,:],template[42,:],template[45,:])))
pts2 = np.float32(template[27:45, :])
# pts2 = np.float32(template[17:35,:])
# pts1 = np.vstack((landmark[27:36,:], landmark[39,:],landmark[42,:],landmark[45,:]))
pts1 = np.float32(landmark[27:45, :])
# pts1 = np.float32(landmark[17:35,:])
tform = tf.SimilarityTransform()
tform.estimate(pts2, pts1)
dst = tf.warp(roi, tform, output_shape=(163, 163))
dst = np.array(dst * 255, dtype=np.uint8)
dst = dst[1:129, 1:129, :]
cv2.imwrite(region_path, dst)
gray = cv2.cvtColor(dst, cv2.COLOR_BGR2GRAY)
# detect faces in the grayscale image
rects = detector(gray, 1)
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
shape, _, _ = normLmarks(shape)
np.save(landmark_path, shape)
lmark = shape.reshape(68, 2)
name = region_path.replace('region.jpg', 'lmark.png')
utils.plot_flmarks(lmark,
name, (-0.2, 0.2), (-0.2, 0.2),
'x',
'y',
figsize=(10, 10))
def crop_image(image_path, gg, last):
image = cv2.imread(image_path)
if gg == 1:
[new_y, new_x, r] = last
roi = image[new_y:new_y + 2 * r, new_x:new_x + 2 * r]
roi = cv2.resize(roi, (256, 256), interpolation=cv2.INTER_AREA)
return roi, last
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
rects = detector(gray, 1)
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
(x, y, w, h) = utils.rect_to_bb(rect)
# cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
# cv2.putText(image, "c", (x - 10, y - 10),
# cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
# for (x, y) in shape:
# cv2.circle(image, (x, y), 1, (0, 0, 255), -1)
# cv2.putText(image, "a", (x , y ),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
center_x = x + int(0.5 * w)
center_y = y + int(0.5 * h)
# cv2.putText(image, "b", (center_x , center_y ),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)
r = int(0.8 * h)
new_x = max(center_x - r, 0)
new_y = max(center_y - r, 0)
# cv2.putText(image, "f", (new_x , new_y ),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 2)
# print (new_x, new_y, r)
roi = image[new_y:new_y + 2 * r, new_x:new_x + 2 * r]
roi = cv2.resize(roi, (256, 256), interpolation=cv2.INTER_AREA)
# scale = 256. / (2 * r)
# shape = ((shape - np.array([new_x,new_y])) * scale)
return roi, [new_y, new_x, r]
def crop_image(image, id = 1, kxy = []):
# kxy = []
if id == 1:
image = cv2.imread(image)
if kxy != []:
[k, x, y] = kxy
roi = image[y - int(0.2 * k):y + int(1.6 * k), x- int(0.4 * k):x + int(1.4 * k)]
roi = cv2.resize(roi, (224,224), interpolation = cv2.INTER_AREA)
return roi, kxy
else:
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
rects = detector(gray, 1)
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
(x, y, w, h) = utils.rect_to_bb(rect)
center_x = x + int(0.5 * w)
center_y = y + int(0.5 * h)
k = min(w, h)
roi = image[y - int(0.2 * k):y + int(1.6 * k), x- int(0.4 * k):x + int(1.4 * k)]
roi = cv2.resize(roi, (224,224), interpolation = cv2.INTER_AREA)
return roi ,[k,x,y]
def getLandmark(self, bounding_box, img):
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
rect = dlib.rectangle(bounding_box[0], bounding_box[1], bounding_box[2], bounding_box[3])
shape = self.predictor(gray, rect)
shape = shape_to_np(shape)
return shape
def generating_demo_landmark_lips(folder='../demo'):
global CC
if not os.path.exists(os.path.join(folder, 'regions')):
os.mkdir(os.path.join(folder, 'regions'))
if not os.path.exists(os.path.join(folder, 'landmark1d')):
os.mkdir(os.path.join(folder, 'landmark1d'))
this = 'obama'
image_txt = os.listdir(os.path.join(folder, 'image', this))
image_txt = sorted(image_txt)
for line in image_txt:
img_path = os.path.join(folder, 'image', this, line)
if not os.path.exists(os.path.join(folder, 'regions', this)):
os.mkdir(os.path.join(folder, 'regions', this))
if not os.path.exists(os.path.join(folder, 'landmark1d', this)):
os.mkdir(os.path.join(folder, 'landmark1d', this))
landmark_path = os.path.join(folder, 'landmark1d', this,
line.replace('jpg', 'npy'))
region_path = os.path.join(folder, 'regions', this, line)
try:
print(img_path)
roi, landmark = crop_image(img_path)
if np.sum(landmark[37:39, 1] - landmark[40:42, 1]) < -9:
# pts2 = np.float32(np.array([template[36],template[45],template[30]]))
template = np.load('../basics/base_68.npy')
else:
template = np.load('../basics/base_68_close.npy')
# pts2 = np.float32(np.vstack((template[27:36,:], template[39,:],template[42,:],template[45,:])))
pts2 = np.float32(template[27:45, :])
# pts2 = np.float32(template[17:35,:])
# pts1 = np.vstack((landmark[27:36,:], landmark[39,:],landmark[42,:],landmark[45,:]))
pts1 = np.float32(landmark[27:45, :])
# pts1 = np.float32(landmark[17:35,:])
tform = tf.SimilarityTransform()
tform.estimate(pts2, pts1)
dst = tf.warp(roi, tform, output_shape=(163, 163))
dst = np.array(dst * 255, dtype=np.uint8)
dst = dst[1:129, 1:129, :]
cv2.imwrite(region_path, dst)
gray = cv2.cvtColor(dst, cv2.COLOR_BGR2GRAY)
# detect faces in the grayscale image
rects = detector(gray, 1)
for (i, rect) in enumerate(rects):
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
np.save(landmark_path, shape)
print(CC)
CC += 1
except:
cv2.imwrite(region_path, dst)
np.save(landmark_path, shape)
continue
print(line)
# loop over the frames from the video stream
while True:
# grab the frame from the threaded video stream, resize it to
# have a maximum width of 400 pixels, and convert it to grayscale
frame = vs.read()
frame = utils.resize(frame, width=400)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# detect faces in the grayscale frame
rects = detector(gray, 0)
# loop over the face detections
for rect in rects:
# determine the facial landmarks for the face region, then
# convert the facial landmark (x, y)-coordinates to a NumPy array
shape = predictor(gray, rect)
shape = utils.shape_to_np(shape)
# loop over the (x, y)-coordinates for the facial landmarks
# and draw them on the image
for (x, y) in shape:
cv2.circle(frame, (x, y), 1, (0, 0, 255), -1)
# show the frame
cv2.imshow("Frame", frame)
key = cv2.waitKey(1) & 0xFF
# if the `q` key was pressed, break from the loop
if key == ord("q"):
break
