github/deepface
1
1
Fork 0
mirror of https://github.com/serengil/deepface.git synced 2025-07-29 14:21:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

Refactor so we can use detect_faces and reshape_face separately

Browse Source 
Now they are separated they can be recombined in the DeepFace.analyse to
allow easier multiple face detection.

We have preprocess_face always return the first face, to ensure its
behaviour doesn't change for other callers.
This commit is contained in:
Roelof Ruis 2021-08-25 10:53:01 +02:00
parent 1a944e7619
commit 50cb600ac5
1 changed files with 23 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
deepface/commons/functions.py
View File

@ -90,7 +90,10 @@ def load_image(img):
return img
def detect_faces(img, detector_backend='opencv', align=True):
def detect_faces(img, enforce_detection=True, detector_backend='opencv', align=True):
# img might be path, base64 or numpy array. Convert it to numpy whatever it is.
img = load_image(img)
img_region = [0, 0, img.shape[0], img.shape[1]]
# ----------------------------------------------
@ -109,6 +112,18 @@ def detect_faces(img, detector_backend='opencv', align=True):
except: # if detected face shape is (0, 0) and alignment cannot be performed, this block will be run
faces = []
if len(faces) == 0:
if not enforce_detection:
raise ValueError(
"No face could be detected. Please confirm that the picture is a face photo or consider "
"to set enforce_detection param to False.")
else:
faces = [(img, [0, 0, img.shape[0], img.shape[1]])] # Set whole image as the detected face
for img, region in faces:
if (img.shape[0] == 0 or img.shape[1] == 0) and enforce_detection:
raise ValueError(f'Detected face shape is {img.shape}, Consider to set enforce_detection argument to False.')
return faces
@ -158,34 +173,19 @@ def normalize_input(img, normalization='base'):
return img
def preprocess_face(img, target_size=(224, 224), grayscale=False, enforce_detection=True, detector_backend='opencv', align=True, allow_multiple=False):
def preprocess_face(img, target_size=(224, 224), grayscale=False, enforce_detection=True, detector_backend='opencv', align=True):
# img might be path, base64 or numpy array. Convert it to numpy whatever it is.
img = load_image(img)
base_img = img.copy()
detected_faces = detect_faces(img=img, detector_backend=detector_backend, align=align)
detected_faces = detect_faces(img=img, enforce_detection=enforce_detection, detector_backend=detector_backend, align=align)
if len(detected_faces) == 0:
if not enforce_detection:
raise ValueError("Face could not be detected. Please confirm that the picture is a face photo or consider "
"to set enforce_detection param to False.")
else:
detected_faces = [(img, [0, 0, img.shape[0], img.shape[1]])]
# only take first detected face
img, region = detected_faces[0]
if img.shape[0] == 0 or img.shape[1] == 0:
img = base_img.copy()
results = []
for img, region in detected_faces:
if img.shape[0] == 0 or img.shape[1] == 0:
if enforce_detection:
raise ValueError(f'Detected face shape is {img.shape}, Consider to set enforce_detection argument to False.')
else: # restore base image
img = base_img.copy()
results.append(reshape_face(img, region, target_size, grayscale))
if not allow_multiple: # if a single result is desired, return
return results[0]
return results
return reshape_face(img, region, target_size, grayscale)
def reshape_face(img, region, target_size=(224, 224), grayscale=False):