def main():
symbol = lightened_moon_feature(num_classes=40, use_fuse=True)
devs = mx.cpu() if args.gpus is None else [mx.gpu(int(i)) for i in args.gpus.split(',')]
_, arg_params, aux_params = mx.model.load_checkpoint(args.model_load_prefix, args.model_load_epoch)
detector = dlib.get_frontal_face_detector()
face_cascade = cv2.CascadeClassifier(args.opencv)
# read img and drat face rect
img = cv2.imread(args.img)
faces = detector(img, 1)
gray = np.zeros(img.shape[0:2])
if len(faces) == 0:
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
opencv_faces = face_cascade.detectMultiScale(gray, 1.3, 5)
for (x,y,w,h) in opencv_faces:
faces.append(dlib.rectangle(int(x), int(y), int(x+w), int(y+h)))
max_face = faces[0]
if len(faces) > 0:
max_face = max(faces, key=lambda rect: rect.width() * rect.height())
for f in faces:
if f == max_face:
cv2.rectangle(img, (f.left(), f.top()), (f.right(), f.bottom()), (0,0,255), 2)
else:
cv2.rectangle(img, (f.left(), f.top()), (f.right(), f.bottom()), (255,0,0), 2)
cv2.imwrite(args.img.replace('jpg', 'png'), img)
# crop face area
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
pad = [0.25, 0.25, 0.25, 0.25] if args.pad is None else args.pad
left = int(max(0, max_face.left() - max_face.width()*float(pad[0])))
top = int(max(0, max_face.top() - max_face.height()*float(pad[1])))
right = int(min(gray.shape[1], max_face.right() + max_face.width()*float(pad[2])))
bottom = int(min(gray.shape[0], max_face.bottom()+max_face.height()*float(pad[3])))
gray = gray[left:right, top:bottom]
gray = cv2.resize(gray, (args.size, args.size))/255.0
img = np.expand_dims(np.expand_dims(gray, axis=0), axis=0)
# get pred
arg_params['data'] = mx.nd.array(img, devs)
exector = symbol.bind(devs, arg_params ,args_grad=None, grad_req="null", aux_states=aux_params)
exector.forward(is_train=False)
exector.outputs[0].wait_to_read()
output = exector.outputs[0].asnumpy()
text = ["5_o_Clock_Shadow","Arched_Eyebrows","Attractive","Bags_Under_Eyes","Bald", "Bangs","Big_Lips","Big_Nose",
"Black_Hair","Blond_Hair","Blurry","Brown_Hair","Bushy_Eyebrows","Chubby","Double_Chin","Eyeglasses","Goatee",
"Gray_Hair", "Heavy_Makeup","High_Cheekbones","Male","Mouth_Slightly_Open","Mustache","Narrow_Eyes","No_Beard",
"Oval_Face","Pale_Skin","Pointy_Nose","Receding_Hairline","Rosy_Cheeks","Sideburns","Smiling","Straight_Hair",
"Wavy_Hair","Wearing_Earrings","Wearing_Hat","Wearing_Lipstick","Wearing_Necklace","Wearing_Necktie","Young"]
pred = np.ones(40)
print("attribution is:")
for i in range(40):
print text[i].rjust(20)+" : \t",
if output[0][i] < 0:
pred[i] = -1
print "No"
else:
pred[i] = 1
print "Yes"
def main():
_, model_args, model_auxs = mx.model.load_checkpoint(
args.model_prefix, args.epoch)
symbol = lightened_moon_feature()
cnt = correct_cnt = 0
with open(args.test_list, 'r') as f:
label = np.ones(40)
pred = np.ones(40)
for line in f.readlines():
# get img and label
line_list = line.strip('\n').split('\t')
img_name = line_list[-1]
logging.info("processing {}, the {}th image".format(
img_name, int(cnt)))
for label_idx in range(1, 41):
label[label_idx - 1] = int(line_list[label_idx])
img = cv2.imread(os.path.join(args.root, img_name), 0) / 255.0
img = cv2.resize(img, (args.size, args.size))
img = np.expand_dims(np.expand_dims(img, axis=0), axis=0)
# get pred
model_args['data'] = mx.nd.array(img, ctx)
exector = symbol.bind(ctx,
model_args,
args_grad=None,
grad_req="null",
aux_states=model_auxs)
exector.forward(is_train=False)
exector.outputs[0].wait_to_read()
output = exector.outputs[0].asnumpy()
for i in range(40):
if output[0][i] < 0:
pred[i] = -1
else:
pred[i] = 1
correct_cnt += (label == pred).sum()
cnt += 1.0
logging.info("acc is:{}".format(correct_cnt / (cnt * 40.0)))
def main():
symbol = lightened_moon_feature(num_classes=40, use_fuse=True)
devs = mx.cpu() if args.gpus is None else [
mx.gpu(int(i)) for i in args.gpus.split(',')
]
_, arg_params, aux_params = mx.model.load_checkpoint(
args.model_load_prefix, args.model_load_epoch)
detector = dlib.get_frontal_face_detector()
face_cascade = cv2.CascadeClassifier(args.opencv)
# read img and drat face rect
img = cv2.imread(args.img)
faces = detector(img, 1)
gray = np.zeros(img.shape[0:2])
if len(faces) == 0:
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
opencv_faces = face_cascade.detectMultiScale(gray, 1.3, 5)
for (x, y, w, h) in opencv_faces:
faces.append(dlib.rectangle(int(x), int(y), int(x + w),
int(y + h)))
max_face = faces[0]
if len(faces) > 0:
max_face = max(faces, key=lambda rect: rect.width() * rect.height())
for f in faces:
if f == max_face:
cv2.rectangle(img, (f.left(), f.top()), (f.right(), f.bottom()),
(0, 0, 255), 2)
else:
cv2.rectangle(img, (f.left(), f.top()), (f.right(), f.bottom()),
(255, 0, 0), 2)
cv2.imwrite(args.img.replace('jpg', 'png'), img)
# crop face area
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
pad = [0.25, 0.25, 0.25, 0.25] if args.pad is None else args.pad
left = int(max(0, max_face.left() - max_face.width() * float(pad[0])))
top = int(max(0, max_face.top() - max_face.height() * float(pad[1])))
right = int(
min(gray.shape[1],
max_face.right() + max_face.width() * float(pad[2])))
bottom = int(
min(gray.shape[0],
max_face.bottom() + max_face.height() * float(pad[3])))
gray = gray[left:right, top:bottom]
gray = cv2.resize(gray, (args.size, args.size)) / 255.0
img = np.expand_dims(np.expand_dims(gray, axis=0), axis=0)
# get pred
arg_params['data'] = mx.nd.array(img, devs)
exector = symbol.bind(devs,
arg_params,
args_grad=None,
grad_req="null",
aux_states=aux_params)
exector.forward(is_train=False)
exector.outputs[0].wait_to_read()
output = exector.outputs[0].asnumpy()
text = [
"5_o_Clock_Shadow", "Arched_Eyebrows", "Attractive", "Bags_Under_Eyes",
"Bald", "Bangs", "Big_Lips", "Big_Nose", "Black_Hair", "Blond_Hair",
"Blurry", "Brown_Hair", "Bushy_Eyebrows", "Chubby", "Double_Chin",
"Eyeglasses", "Goatee", "Gray_Hair", "Heavy_Makeup", "High_Cheekbones",
"Male", "Mouth_Slightly_Open", "Mustache", "Narrow_Eyes", "No_Beard",
"Oval_Face", "Pale_Skin", "Pointy_Nose", "Receding_Hairline",
"Rosy_Cheeks", "Sideburns", "Smiling", "Straight_Hair", "Wavy_Hair",
"Wearing_Earrings", "Wearing_Hat", "Wearing_Lipstick",
"Wearing_Necklace", "Wearing_Necktie", "Young"
]
pred = np.ones(40)
print("attribution is:")
for i in range(40):
print text[i].rjust(20) + " : \t",
if output[0][i] < 0:
pred[i] = -1
print "No"
else:
pred[i] = 1
print "Yes"
def main(args):
#MOON feature extractor; not sure how to make this a modular component
symbol = lightened_moon_feature(num_classes=40, use_fuse=True)
#the detector passed in from the command line; requires files in facenet/data/
detector = align_dlib.AlignDlib(
os.path.expanduser(args.dlib_face_predictor))
landmarkIndices = align_dlib.AlignDlib.OUTER_EYES_AND_NOSE
if args.landmarkIndices is not None:
landmarkIndices = args.landmarkIndices
video = cv2.VideoCapture(args.input_video)
devs = mx.cpu()
#begin to iterate over the frames and process them
ret, frame = video.read()
#a list of dictionaries containing face_output for each frame
total_output = []
#maps encoding matrix to id number
known_faces_dict = dict()
known_faces_encoding = []
id_count = 0
while ret is True:
face_boxes = detector.getAllFaceBoundingBoxes(frame)
id_attr, known_faces_dict, known_faces_encoding, id_count = processFrame(
args, frame, known_faces_dict, known_faces_encoding, id_count,
symbol, detector, landmarkIndices, devs, face_boxes)
total_output.append(id_attr)
ret, frame = video.read()
#==========CONVERT TO JSON FILE===============
print('done processing; converting to json')
#print(total_output)
#ith element represents the ith frame
frame_num = 0
json_output = '{\r\n"video":\r\n{\r\n"frames":\r\n[\r\n'
for frame_info in total_output:
#begin the num-faces entry
json_output += '{\r\n"num": ' + str(frame_num) + ',\r\n'
if len(frame_info.keys()) == 0:
#if this still isnt valid, try doing "faces": 0 and closing the field
# remove last occurrence of comma
k = json_output.rfind(',')
json_output = json_output[:k] + json_output[k + 1:]
json_output += '},\r\n' #close the num-faces entry; no faces field
frame_num += 1
continue
json_output += '"faces":\r\n[\r\n'
# process the face information in frame_info in a loop
for face in frame_info.keys():
#get actual content, which is a list
#content shouldnt ever be empty, because there exists a key
#TODO may be a bug bc of this assumption
content = frame_info[face]
pid = content[0]
#check if content is length > 1
#there may be an individual with 0 yes-attributes
if len(content) == 3:
#attributes will contain the topleft,bottomright coordinates,
#followed by the attributes themselves
attributes = content[1:len(content) - 1]
attributes.extend(['Negatives'])
d = {pid: attributes} #looks like 0:[]
json_output += json.dumps(d) + ',\r\n'
continue
#attributes will contain the topleft, bottomright coordinates
#followed by the attributes themselves
attributes = content[1:len(content) - 1]
d = {pid: attributes}
#now we have the proper split
json_output += json.dumps(d) + ',\r\n'
#outside of loop
# remove last occurrence of comma
k = json_output.rfind(',')
json_output = json_output[:k] + json_output[k + 1:]
json_output += ']\r\n' #close the faces array
json_output += '},\r\n' #close the num-faces entry
frame_num += 1
# remove last occurrence of comma
k = json_output.rfind(',')
json_output = json_output[:k] + json_output[k + 1:]
json_output += '\r\n]\r\n}\r\n}'
d = json.loads(json_output)
json_output = json.dumps(d, indent=4, separators=(',', ': '))
#write out to file
print('done converting to json; writing to file')
f = open('output.json', 'wb')
f.write(json_output)
f.close()
print('done!')
