github/deepface
1
1
Fork 0
mirror of https://github.com/serengil/deepface.git synced 2025-06-07 12:05:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

simplified code

Browse Source
This commit is contained in:
Sefik Ilkin Serengil 2022-01-19 12:07:57 +03:00 committed by GitHub
parent 3485c348bc
commit 5171484957
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 33 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

94
deepface/detectors/MediapipeWrapper.py
View File

@ -1,77 +1,49 @@
from deepface.detectors import FaceDetector from deepface.detectors import FaceDetector
# Link - https://google.github.io/mediapipe/solutions/face_detection # Link - https://google.github.io/mediapipe/solutions/face_detection
def build_model(): def build_model():
import mediapipe as mp import mediapipe as mp #this is not a must dependency. do not import it in the global level.
mp_face_detection = mp.solutions.face_detection mp_face_detection = mp.solutions.face_detection
# Build a face detector face_detection = mp_face_detection.FaceDetection( min_detection_confidence=0.7)
# min_detection_confidence - "A filter to analyse the training photographs"
face_detection = mp_face_detection.FaceDetection( min_detection_confidence=0.6)
return face_detection return face_detection
def detect_face(face_detector, img, align=True): def detect_face(face_detector, img, align = True):
import mediapipe as mp import mediapipe as mp #this is not a must dependency. do not import it in the global level.
import re
#mp_face_detection = mp.solutions.face_detection
#mp_drawing = mp.solutions.drawing_utils
resp = [] resp = []
img_width = img.shape[1]; img_height = img.shape[0]
results = face_detector.process(img) results = face_detector.process(img)
original_size = img.shape
target_size = (300, 300) if results.detections:
# First face , than eye
#print(results.detections)
if results.detections:
for detection in results.detections: for detection in results.detections:
#mp_drawing.draw_detection(img, detection)
#print(detection) confidence = detection.score
# detected_face is the cropped image that is then passed forward to the Regognizer
''' bounding_box = detection.location_data.relative_bounding_box
DETECTION - landmarks = detection.location_data.relative_keypoints
Collection of detected faces, where each face is represented as a detection proto message that contains
a bounding box and 6 key points (right eye, left eye, nose tip, mouth center, right ear tragion, and left x = int(bounding_box.xmin * img_width)
ear tragion). The bounding box is composed of xmin and width (both normalized to [0.0, 1.0] by the w = int(bounding_box.width * img_width)
image width) and ymin and height (both normalized to [0.0, 1.0] by the image height). Each key point y = int(bounding_box.ymin * img_height)
is composed of x and y, which are normalized to [0.0, 1.0] by the image width and height h = int(bounding_box.height * img_height)
respectively.
''' right_eye = (int(landmarks[0].x * img_width), int(landmarks[0].y * img_height))
# Bounding Box left_eye = (int(landmarks[1].x * img_width), int(landmarks[1].y * img_height))
x = re.findall('xmin: (..*)',str(detection)) #nose = (int(landmarks[2].x * img_width), int(landmarks[2].y * img_height))
y = re.findall('ymin: (..*)',str(detection)) #mouth = (int(landmarks[3].x * img_width), int(landmarks[3].y * img_height))
h = re.findall('height: (..*)',str(detection)) #right_ear = (int(landmarks[4].x * img_width), int(landmarks[4].y * img_height))
w = re.findall('width: (..*)',str(detection)) #left_ear = (int(landmarks[5].x * img_width), int(landmarks[5].y * img_height))
# Eye Locations
reye_x = re.findall('x: (..*)',str(detection))[0] if x > 0 and y > 0:
leye_x = re.findall('x: (..*)',str(detection))[1] detected_face = img[y:y+h, x:x+w]
reye_y = re.findall('y: (..*)', str(detection))[0] img_region = [x, y, w, h]
leye_y = re.findall('y: (..*)', str(detection))[1]
# Detections are normalized by the mediapipe API, thus they need to be multiplied
# Extra tweaking done to improve accuracy
x = (float(x[0]) * original_size[1])
y = (float(y[0]) * original_size[0]-15)
h = (float(h[0]) * original_size[0]+10)
w = (float(w[0]) * original_size[1]+10)
reye_x = (float(reye_x) * original_size[1])
leye_x = (float(leye_x) * original_size[1])
reye_y = (float(reye_y) * original_size[0])
leye_y = (float(leye_y) * original_size[0])
if float(x) and float(y) > 0:
detected_face = img[int(y):int(y + h), int(x):int(x + w)]
img_region = [int(x), int(y), int(w), int(h)]
if align: if align:
left_eye=(leye_x,leye_y)
right_eye=(reye_x,reye_y)
#print(left_eye)
#print(right_eye)
detected_face = FaceDetector.alignment_procedure(detected_face, left_eye, right_eye) detected_face = FaceDetector.alignment_procedure(detected_face, left_eye, right_eye)
resp.append((detected_face,img_region)) resp.append((detected_face,img_region))
else:
continue
#print("Yahoo")
return resp return resp
#face_detector = FaceDetector.build_model('mediapipe')