def main():
if cv2gpu.is_cuda_compatible():
cv2gpu.init_gpu_detector(cascade_file_gpu)
else:
cv2gpu.init_cpu_detector(cascade_file_cpu)
image = cv2.imread(image_file, cv2.IMREAD_COLOR)
dims = image.shape
image = image.ravel()
# Run once it initialize gpu with instructions
faces = cv2gpu.find_faces(dims, image)
start = datetime.datetime.now()
faces = cv2gpu.find_faces(dims, image)
image = cv2.imread(image_file)
for (x, y, w, h) in faces:
print('X: {x}, Y: {y}, width: {w}, height: {h}'.format(x=x, y=y, w=w, h=h))
#cv2.rectangle(image, (x, y), (x+w, y+h), (0, 255, 0), 2)
#cv2.imshow('faces', image)
#cv2.waitKey(0)
print('Time processed (s): ' + str(datetime.datetime.now() - start))
def separate_images(image_obj):
size_file_max = 200000
def prune_func_gpu(image_path):
if getsize(image_path) < size_file_max:
faces = cv2gpu.find_faces(image_path)
if faces:
return len(faces) > 0
return False
def prune_func_cpu(image_path):
if getsize(image_path) < size_file_max:
image = cv2.imread(image_path, 0)
faces = cascade.detectMultiScale(image)
if faces:
return len(faces) > 0
return False
if not isfile(cascade_gpu_path) or not isfile(cascade_cpu_path):
print '{} or {} is not a file'.format(cascade_gpu_path, cascade_cpu_path)
sys.exit(1)
prune_func = None
if cv2gpu_imported:
prune_func = prune_func_gpu
if cv2gpu.is_cuda_compatible():
print 'Using GPU detector'
cv2gpu.init_gpu_detector(cascade_gpu_path)
else:
print 'Using CPU detector'
cv2gpu.init_cpu_detector(cascade_cpu_path)
else:
prune_func = prune_func_cpu
cascade = cv2.CascadeClassifier(cascade_cpu_path)
image_num_max = count_images(image_obj)
image_num_curr = 0
for tld in image_obj:
out_dir = join(processed_dir, tld)
if not isdir(out_dir):
makedirs(out_dir)
images = image_obj[tld]
for url in images:
path_orig = images[url]
path_new = join(out_dir, basename(path_orig))
print 'Image {} / {}, ({})'.format(image_num_curr, image_num_max, getsize(path_orig))
image_num_curr += 1
if not prune_func(path_orig):
continue
if not isfile(path_new):
copy2(path_orig, path_new)
images[url] = path_new
return image_obj
def main():
if cv2gpu.is_cuda_compatible():
cv2gpu.init_gpu_detector(cascade_file_gpu)
else:
cv2gpu.init_cpu_detector(cascade_file_cpu)
faces = cv2gpu.find_faces(image_file)
image = cv2.imread(image_file)
for (x, y, w, h) in faces:
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.imshow('faces', image)
cv2.waitKey(0)
def main():
if cv2gpu.is_cuda_compatible():
cv2gpu.init_gpu_detector(cascade_file_gpu)
else:
cv2gpu.init_cpu_detector(cascade_file_cpu)
faces = cv2gpu.find_faces(image_file)
image = cv2.imread(image_file)
for (x, y, w, h) in faces:
cv2.rectangle(image, (x, y), (x+w, y+h), (0, 255, 0), 2)
cv2.imshow('faces', image)
cv2.waitKey(0)
def create_detector(gpu_or_cpu, cascade_cpu, cascade_gpu):
if not os.path.isfile(cascade_cpu) or not os.path.isfile(cascade_gpu):
sys.exit('{} or {} is not a file'.format(cascade_cpu, cascade_gpu))
if cv2gpu.is_cuda_compatible() and gpu_or_cpu == 'gpu':
sys.stdout.write('Using GPU CascadeClassifier\n')
cv2gpu.init_gpu_detector(cascade_gpu)
else:
sys.stdout.write('Using CPU CascadeClassifier\n')
cv2gpu.init_cpu_detector(cascade_cpu)
class Detector(object):
def detect(self, image_path):
return cv2gpu.find_faces(image_path)
return Detector()
net = openface.TorchNeuralNet(args.networkModel,
imgDim=args.imgDim,
cuda=args.cuda)
face_cascade = cv2.CascadeClassifier(
'/root/openface/data/lbpcascades/lbpcascade_frontalface_improved.xml')
#face_cascade = cv2.CascadeClassifier('/root/openface/data/lbpcascades/lbpcascade_profileface.xml')
if face_cascade.empty():
print("Can not load face cascade classifier xml!")
if cv2gpu.is_cuda_compatible():
cv2gpu.init_gpu_detector(
'/root/openface/data/haarcascade_frontalface_default_cuda.xml')
#cv2gpu.init_gpu_detector('/root/openface/data/lbpcascades/lbpcascade_frontalface_improved.xml')
print("Using gpu haarcascade!")
else:
cv2gpu.init_cpu_detector(
'/root/openface/data/haarcascade_frontalface_default.xml')
print("Using cpu haarcascade only!")
if args.verbose:
print("Loading the dlib and OpenFace models took {} seconds.".format(
time.time() - start))
start = time.time()
def round_to_1(x):
if x == 0:
return 0
else:
return round(x, -int(floor(log10(abs(x)))))
def mapper_init(self):
def load_from_small_dataset(colorferet_small_dir):
face_labels_str = {
'Black-or-African-American': 0,
'Asian': 1,
'Asian-Middle-Eastern': 2,
'Hispanic': 3,
'Native-American': 4,
'Other': 5,
'Pacific-Islander': 6,
'White': 7
}
images = []
labels = []
for face_label_str in face_labels_str:
face_label_num = face_labels_str[face_label_str]
image_dir = os.path.join(colorferet_small_dir, face_label_str)
image_files = [
os.path.join(image_dir, image_file)
for image_file in os.listdir(image_dir)
]
images_tmp = [
cv2.resize(cv2.imread(image_file, 0), (256, 256))
for image_file in image_files
if image_file.split('.')[-1] == 'png'
]
labels_tmp = [face_label_num] * len(images_tmp)
images.extend(images_tmp)
labels.extend(labels_tmp)
return images, labels
self.video_dir = jobconf_from_env('job.settings.video_dir')
self.output_dir = os.path.join(
jobconf_from_env('mapreduce.task.output.dir'), 'faces')
self.opencv_version = int(cv2.__version__.split('.')[0])
if self.opencv_version == 2:
self.recognizer = cv2.createLBPHFaceRecognizer()
# self.recognizer = cv2.createFisherFaceRecognizer()
# self.recognizer = cv2.createEigenFaceRecognizer()
elif self.opencv_version == 3:
self.recognizer = cv2.face.createLBPHFaceRecognizer()
# self.recognizer = cv2.face.createFisherFaceRecognizer()
# self.recognizer = cv2.face.createEigenFaceRecognizer()
if cv2gpu.is_cuda_compatible():
sys.stderr.write('Using GPU CascadeClassifier')
cv2gpu.init_gpu_detector(
jobconf_from_env('job.settings.cascade_gpu'))
else:
sys.stderr.write('Using CPU CascadeClassifier')
cv2gpu.init_cpu_detector(
jobconf_from_env('job.settings.cascade_cpu'))
images, labels = load_from_small_dataset(
jobconf_from_env('job.settings.colorferet'))
self.recognizer.train(images, numpy.array(labels))
if not os.path.exists(self.output_dir):
os.makedirs(self.output_dir)
self.face_labels_num = {
0: 'Black-or-African-American',
1: 'Asian',
2: 'Asian-Middle-Eastern',
3: 'Hispanic',
4: 'Native-American',
5: 'Other',
6: 'Pacific-Islander',
7: 'White'
}
self.write_results = False
args = parser.parse_args()
print("Initializing face detector...")
if args.detector == 'hog':
detector = dlib.get_frontal_face_detector()
elif args.detector == 'cnn':
detector = dlib.cnn_face_detection_model_v1(args.cnn_model_path)
elif args.detector == 'opencv':
detector = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
elif args.detector == 'cv2gpu':
import cv2gpu
if cv2gpu.is_cuda_compatible():
cv2gpu.init_gpu_detector('haarcascade_frontalface_default_cuda.xml')
else:
cv2gpu.init_cpu_detector('haarcascade_frontalface_default.xml')
else:
assert False, "Unknown detector " + args.detector
predictor = dlib.shape_predictor(args.shape_predictor_path)
facerec = dlib.face_recognition_model_v1(args.face_rec_model_path)
print("Loading feature vectors...")
vecs = np.load(args.vecs_file)
nn = NearestNeighbors(n_neighbors=args.n_neighbors, algorithm=args.algorithm)
nn.fit(vecs)
print("Loading image metadata...")
with open(args.data_file) as csvfile:
reader = csv.DictReader(
csvfile, fieldnames=['file', 'name', 'left', 'top', 'right', 'bottom'])
data = list(reader)
def mapper_init(self):
def load_from_small_dataset(colorferet_small_dir):
face_labels_str = {
'Black-or-African-American': 0,
'Asian': 1,
'Asian-Middle-Eastern': 2,
'Hispanic': 3,
'Native-American': 4,
'Other': 5,
'Pacific-Islander': 6,
'White': 7
}
images = []
labels = []
for face_label_str in face_labels_str:
face_label_num = face_labels_str[face_label_str]
image_dir = os.path.join(colorferet_small_dir, face_label_str)
image_files = [os.path.join(image_dir, image_file) for image_file in os.listdir(image_dir)]
images_tmp = [cv2.resize(cv2.imread(image_file, 0), (256, 256)) for image_file in image_files if image_file.split('.')[-1] == 'png']
labels_tmp = [face_label_num] * len(images_tmp)
images.extend(images_tmp)
labels.extend(labels_tmp)
return images, labels
self.video_dir = jobconf_from_env('job.settings.video_dir')
self.output_dir = os.path.join(jobconf_from_env('mapreduce.task.output.dir'), 'faces')
self.opencv_version = int(cv2.__version__.split('.')[0])
if self.opencv_version == 2:
self.recognizer = cv2.createLBPHFaceRecognizer()
# self.recognizer = cv2.createFisherFaceRecognizer()
# self.recognizer = cv2.createEigenFaceRecognizer()
elif self.opencv_version == 3:
self.recognizer = cv2.face.createLBPHFaceRecognizer()
# self.recognizer = cv2.face.createFisherFaceRecognizer()
# self.recognizer = cv2.face.createEigenFaceRecognizer()
if cv2gpu.is_cuda_compatible():
sys.stderr.write('Using GPU CascadeClassifier')
cv2gpu.init_gpu_detector(jobconf_from_env('job.settings.cascade_gpu'))
else:
sys.stderr.write('Using CPU CascadeClassifier')
cv2gpu.init_cpu_detector(jobconf_from_env('job.settings.cascade_cpu'))
images, labels = load_from_small_dataset(jobconf_from_env('job.settings.colorferet'))
self.recognizer.train(images, numpy.array(labels))
if not os.path.exists(self.output_dir):
os.makedirs(self.output_dir)
self.face_labels_num = {
0: 'Black-or-African-American',
1: 'Asian',
2: 'Asian-Middle-Eastern',
3: 'Hispanic',
4: 'Native-American',
5: 'Other',
6: 'Pacific-Islander',
7: 'White'
}
self.write_results = False
