def __init__(self):
self.labels = ["male", "female"]
self.mean = (78.4263377603, 87.7689143744, 114.895847746)
self.f_detector = FDetector()
self.web_util = web()
self.f_detector.set_detector("hog")
self.__set_detector()
from visionlib.face.detection import FDetector
import cv2
import argparse
# Configre the parser.
parser = argparse.ArgumentParser()
parser.add_argument("img_path", help="Path to image")
parser.add_argument("--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool)
args = parser.parse_args()
# Instantiating the required classes.
detector = FDetector()
# Read the image
img = cv2.imread(args.img_path)
detector.set_detector("hog")
# Apply face detection and show image
d_img, boxes, conf = detector.detect_face(img,
show=True,
enable_gpu=args.enable_gpu)
for box, conf in zip(boxes, conf):
print(box, conf)
"--face-model",
help="Model to use for detecting face",
dest="face_model",
default="hog",
)
parser.add_argument(
"--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool,
)
args = parser.parse_args()
# Instantiating the required class
fdetector = FDetector()
kdetector = KDetector()
# Read the image
img = cv2.imread(args.img_path)
# Set the detectors
kdetector.set_detector(args.key_model)
fdetector.set_detector(args.face_model)
# Detect face and the points
fimg, boxes, conf = fdetector.detect_face(img, enable_gpu=True)
points = kdetector.detect_keypoints(img, rects=boxes)
# Draw the points
pimg = kdetector.draw_points(fimg, points, show=True)
from visionlib.face.detection import FDetector
import cv2
import argparse
# Configre the parser.
parser = argparse.ArgumentParser()
parser.add_argument("vid_path", help="Path to image")
parser.add_argument("--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool)
args = parser.parse_args()
# Instantiating the required classes.
detector = FDetector()
detector.set_detector("dnn")
# Read the video and apply face detection.
detection = detector.vdetect_face(args.vid_path,
show=True,
enable_gpu=args.enable_gpu)
for img, box, conf in detection:
print(box, conf)
def setUp(self):
self.detectors = ["mtcnn", "haar", "hog", "dnn"]
self.face_detector = FDetector()
self.group_img_path = "assets/face.jpg"
self.test_img = cv2.imread(self.group_img_path)
logging.basicConfig(level=logging.INFO)
class GDetector:
"""
This class contains all functions to detect gender of a given face
. . .
Methods:
detect_gender():
Used to detect gender from an face. Returns Predicted Gender and
confidence.
"""
def __init__(self):
self.labels = ["male", "female"]
self.mean = (78.4263377603, 87.7689143744, 114.895847746)
self.f_detector = FDetector()
self.web_util = web()
self.f_detector.set_detector("hog")
self.__set_detector()
def __set_detector(self):
model_name = 'gender_net.caffemodel'
model_url = 'https://github.com/arunponnusamy/cvlib-files/releases/download/v0.1/gender_net.caffemodel'
proto_name = 'gender_deploy.prototxt'
proto_url = 'https://download.cvlib.net/config/gender_detection/gender_deploy.prototxt'
self.model = self.web_util.download_file(model_url, model_name)
self.proto = self.web_util.download_file(proto_url, proto_name)
self.model = cv2.dnn.readNet(self.proto, self.model)
def detect_gender(self, img=None, enable_gpu=False):
'''This method is used to detect gender from an image.
Args:
img (numpy array)
This argument must the output which similar to
opencv's imread method's output.
enable_gpu (bool) :
Set to True if You want to use gpu for prediction.
Returns:
str :
Returns the predicted gender.
int :
Returns the confidence for the predicted gender.
'''
if img != []:
blob = cv2.dnn.blobFromImage(img,
1.0, (227, 227),
self.mean,
swapRB=False)
if enable_gpu:
self.model.setPreferableBackend(cv2.dnn.DNN_BACKEND_CUDA)
self.model.setPreferableTarget(cv2.dnn.DNN_TARGET_CUDA)
self.model.setInput(blob)
preds = self.model.forward()
preds = preds[0]
b_pred = self.labels[np.argmax(preds)]
b_confidence = preds[np.argmax(preds)]
return (b_pred, b_confidence)
else:
print("img is none")
from visionlib.gender.detection import GDetector
from visionlib.utils.imgutils import Image
import cv2
import argparse
# Configre the parser.
parser = argparse.ArgumentParser()
parser.add_argument("img_path", help="Path to image")
parser.add_argument("--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool)
args = parser.parse_args()
# Instantiating the required classes.
Fdetector = FDetector()
Gdetector = GDetector()
im_utils = Image()
# Read the image
img = cv2.imread(args.img_path)
Fdetector.set_detector("hog")
# Detect the Faces
d_img, boxes, conf = Fdetector.detect_face(img)
for box in boxes:
# Get the face by cropping
c_img = im_utils.crop(d_img, box)
# Apply Gender Detection
gender = Gdetector.detect_gender(c_img, enable_gpu=args.enable_gpu)
# format the label
label = "{}: {:.2f}%".format(gender[0], gender[1] * 100)
# Put padding for rendering the label
help="The model to use for detection",
dest="model")
parser.add_argument("--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool)
parser.add_argument("--cam",
help="Set to True if you are using webcam",
dest="cam",
default=False,
type=bool)
args = parser.parse_args()
# Instantiating the required classes.
detector = FDetector()
Gdetector = GDetector()
im_utils = Image()
detector.set_detector(args.model)
# Read the video
detection = detector.vdetect_face(args.vid_path, cam=args.cam)
run = True
while run:
# Get frames and bounding boxes of faces
for img, boxes, conf in detection:
for box in boxes:
# Get the face by cropping
c_img = im_utils.crop(img, box)
# Apply Gender Detection
gender = Gdetector.detect_gender(c_img, enable_gpu=args.enable_gpu)
from visionlib.utils.imgutils import Image
import cv2
import argparse
# Configre the parser.
parser = argparse.ArgumentParser()
parser.add_argument("vid_path", help="Path to image")
parser.add_argument("--enable-gpu",
help="Set to true to enable gpu support",
dest="enable_gpu",
default=False,
type=bool)
args = parser.parse_args()
# Instantiating the required classes.
detector = FDetector()
Gdetector = GDetector()
im_utils = Image()
detector.set_detector("dnn")
# Read the video
detection = detector.vdetect_face(args.vid_path)
run = True
while run:
# Get frames and bounding boxes of faces
for img, boxes, conf in detection:
for box in boxes:
# Get the face by cropping
c_img = im_utils.crop(img, box)
# Apply Gender Detection
gender = Gdetector.detect_gender(c_img, enable_gpu=args.enable_gpu)