def detect_one_face_align(img):
crop_size = 112
scale = crop_size / 112.
reference = get_reference_facial_points(default_square=True) * scale
#img = Image.open(os.path.join(image_path))
landmarks = []
bounding_boxes = []
try:
bounding_boxes, landmarks = detect_faces(img)
except Exception as e:
print(e)
if len(
landmarks
) == 0: # If the landmarks cannot be detected, the img will be discarded
return None
#print(landmarks)
facial5point = [[landmarks[0][j], landmarks[0][j + 5]] for j in range(5)]
#print(facial5points[i])
warped_face = warp_and_crop_face(np.array(img),
facial5point,
reference,
crop_size=(crop_size, crop_size))
img_warped = Image.fromarray(warped_face)
# img_warped.save("test.jpg")
# img_warped.show()
return img_warped, bounding_boxes
def detect_face_align(img):
crop_size = 112 #因为backbone的input_size要[112,112]or[224,224]
scale = crop_size / 112.
reference = get_reference_facial_points(default_square=True) * scale
#img = Image.open(os.path.join(image_path))
bounding_boxes = []
warped_face = []
img_warped = []
facial5point = []
landmarks = []
try:
bounding_boxes, landmarks = detect_faces(img)
except Exception as e:
print(e)
if len(
landmarks
) == 0: # If the landmarks cannot be detected, the img will be discarded
return img_warped, bounding_boxes
#print(landmarks)
for i in range(len(landmarks)):
facial5point.append([[landmarks[i][j], landmarks[i][j + 5]]
for j in range(5)])
#print(facial5points[i])
warped_face.append(
warp_and_crop_face(np.array(img),
facial5point[i],
reference,
crop_size=(crop_size, crop_size)))
img_warped.append(Image.fromarray(warped_face[i]))
# img_warped.save("test.jpg")
# img_warped.show()
return img_warped, bounding_boxes
def build(self, image):
#1.add face detection
bounding_boxes, landmarks = detect_faces(image)
if len(bounding_boxes) == 0:
return None
#2.add face align
wrapedfaces = self.warpface(landmarks,image)
#3.compute face embedding
return self.embedding(wrapedfaces)
def main(img_path):
img = Image.open(img_path) # modify the image path to yours
det_models = load_detect_faces_models()
bounding_boxes, landmarks = detect_faces(
det_models,
img) # detect bboxes and landmarks for all faces in the image
viz_img = show_results(img, bounding_boxes,
landmarks) # visualize the results
viz_img.save('out.jpg')
face_id_model = load_face_id_model()
for bbox_idx, bbox in enumerate(bounding_boxes):
tl_x, tl_y, br_x, br_y, prob = bbox.astype(int)
cropped_img = np.array(img)[tl_y:br_y, tl_x:br_x, :]
Image.fromarray(cropped_img).save('bbox-{}.jpg'.format(bbox_idx))
features = extract_feature_for_img(img=cropped_img,
backbone=face_id_model)
print(features)
def align_(img, crop_size, model_root, filter=False):
# check if img is PIL
if type(img) == np.ndarray:
img = Image.fromarray(img)
# path to detector models
op = "{}/onet.npy".format(model_root)
pp = "{}/pnet.npy".format(model_root)
rp = "{}/rnet.npy".format(model_root)
# settings
scale = crop_size / 112.
reference = get_reference_facial_points(default_square=True) * scale
try: # Handle exception
_, landmarks = detect_faces(img, ppath=pp, opath=op, rpath=rp)
except Exception:
print("Image is discarded due to exception!")
return 4
if len(
landmarks
) == 0: # If the landmarks cannot be detected, the img will be discarded
print("Image is discarded due to non-detected landmarks!")
return 4
if filter:
return True
else:
facial5points = [[landmarks[0][j], landmarks[0][j + 5]]
for j in range(5)]
warped_face = warp_and_crop_face(np.array(img),
facial5points,
reference,
crop_size=(crop_size, crop_size))
img_warped = Image.fromarray(warped_face)
return img_warped
source_root = args.source_root # specify your source dir
dest_root = args.dest_root # specify your destination dir
crop_size = args.crop_size # specify size of aligned faces, align and crop with padding
scale = crop_size / 112.
reference = get_reference_facial_points(default_square = True) * scale
cwd = os.getcwd() # delete '.DS_Store' existed in the source_root
os.chdir(source_root)
os.system("find . -name '*.DS_Store' -type f -delete")
os.chdir(cwd)
if not os.path.isdir(dest_root):
os.mkdir(dest_root)
for image_name in tqdm(glob("{}/*".format(source_root))):
print("Processing\t{}".format(image_name))
img = Image.open(image_name)
try: # Handle exception
_, landmarks = detect_faces(img)
except Exception:
print("{} is discarded due to exception!".format(image_name))
continue
if len(landmarks) == 0: # If the landmarks cannot be detected, the img will be discarded
print("{} is discarded due to non-detected landmarks!".format(image_name))
continue
facial5points = [[landmarks[0][j], landmarks[0][j + 5]] for j in range(5)]
warped_face = warp_and_crop_face(np.array(img), facial5points, reference, crop_size=(crop_size, crop_size))
img_warped = Image.fromarray(warped_face)
img_warped.save(os.path.join(dest_root, image_name.split('.')[0] + '.jpg'))
import numpy as np
import torch
from PIL import Image
from models import IR_50
from align.align_trans import warp_and_crop_face, get_reference_facial_points
from align.detector import detect_faces
from align.visualization_utils import show_results
img = Image.open('testImages/stp.jpeg') # modify the image path to yours
bounding_boxes, landmarks = detect_faces(img) # detect bboxes and landmarks for all faces in the image
image = show_results(img, bounding_boxes, landmarks) # visualize the results
crop_size = 112 # specify size of aligned faces, align and crop with padding
scale = crop_size / 112.
reference = get_reference_facial_points(default_square=True) * scale
model = IR_50((112, 112)).eval()
model.load_state_dict(torch.load('backbone_ir50_ms1m_epoch120.pth'))
for landmark in landmarks:
facial5points = [[landmark[j], landmark[j + 5]] for j in range(5)]
warped_face = warp_and_crop_face(np.array(img), facial5points, reference, crop_size=(112, 112))
img_warped = torch.Tensor(warped_face).unsqueeze(0).transpose(1, 3)
out = model(img_warped).detach().cpu().numpy()
norm = np.linalg.norm(out, axis=1)
out = out / norm
i = 5
def main(request):
"""Displays the main page
Arguments:
request {request} -- GET or POST request
"""
# SHOW MAIN PAGE
if request.method == 'GET':
context = {
'title': 'Main',
}
return render(request, 'index.html', context)
# GOT NEW PHOTO
elif request.method == 'POST':
# INITIAL SEARCH
if request.POST.get('refine') == "False":
image = request.FILES.get('photo')
image_type = image.name.split('.')[-1] #png/jpg/jpeg
now = datetime.datetime.now()
image_path = f'{now.day}{now.month}{now.year}_{now.hour}:{now.minute}:{now.second}.{image_type}'
full_path = os.path.join(MEDIA_PATH, image_path)
with open(full_path, 'wb+') as destination:
destination.write(image.read())
img = Image.open(full_path)
_, landmarks = detect_faces(img) #TODO: change onet/rnet/pnet path
if landmarks == []:
# there are no faces on the photo
# TODO: send message
return redirect('Main Page')
count = landmarks.shape[0]
if count == 1:
img = align_face(img, landmarks[0]) # cropped aligned face, ready for search
D, I = search(img) # distances and indexes
results_dict = results(D,I)
context = {
'title': 'Search Results',
'results_dict': results_dict
}
return render(request, 'results.html', context)
else:
face_urls = refine_face(img, landmarks, image_path)
context = {
'title': 'Choose face',
'faces_list': face_urls,
}
return render(request, 'refine.html', context)
# SEARCH AFTER REFINING
elif request.POST.get('refine') == "True":
image_path = request.POST.get('imagecrop') # number selected face
full_path = os.path.join(MEDIA_PATH, image_path)
img = Image.open(full_path) # cropped aligned face, ready for search
D, I = search(img) # distances and indexes
results_dict = results(D,I)
context = {
'title': 'Search Results',
'results_dict': results_dict
}
return render(request, 'results.html', context)
cv2.setWindowProperty(window, cv2.WND_PROP_FULLSCREEN, cv2.WINDOW_FULLSCREEN)
while (True):
#Capture frame
ret, img = capture.read()
# Get frameRate
frameRate_counter += 1
if (time.time() - frameRate_start_time) > frameRate_refresh:
frameRate_fps = frameRate_counter / (time.time() -
frameRate_start_time)
frameRate_counter = 0
frameRate_start_time = time.time()
# Faces location
bounding_boxes, landmarks = detector.detect_faces(Image.fromarray(img))
for i, box in enumerate(bounding_boxes):
# Set color
color = colors.astype('int')[i % 10].tolist()
# Copy just the face
box = box.astype('int')
face = img[box[1]:box[3], box[0]:box[2]]
# Build the face frame
corner_top_left = (int(box[0]), int(box[1]))
corner_buttom_right = (int(box[2]), int(box[3]))
cv2.rectangle(img, corner_top_left, corner_buttom_right, color, 2)
with torch.no_grad():
# Fake batch
face = prepare_face(face)
