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

Merge pull request #801 from AnthraX1/fixyunet

Browse Source 
Fix yunet return negative coordinates
This commit is contained in:
Sefik Ilkin Serengil 2023-07-26 12:06:06 +01:00 committed by GitHub
commit 95d0eeddc6
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 21 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
deepface/detectors/YunetWrapper.py
View File

@ -6,10 +6,8 @@ from deepface.commons import functions
def build_model(): def build_model():
url = ( # pylint: disable=C0301
"https://github.com/opencv/opencv_zoo/raw/main/models/" url = "https://github.com/opencv/opencv_zoo/raw/main/models/face_detection_yunet/face_detection_yunet_2023mar.onnx"
+ "face_detection_yunet/face_detection_yunet_2023mar.onnx"
)
file_name = "face_detection_yunet_2023mar.onnx" file_name = "face_detection_yunet_2023mar.onnx"
home = functions.get_deepface_home() home = functions.get_deepface_home()
if os.path.isfile(home + f"/.deepface/weights/{file_name}") is False: if os.path.isfile(home + f"/.deepface/weights/{file_name}") is False:
@ -44,17 +42,26 @@ def detect_face(detector, image, align=True, score_threshold=0.9):
if faces is None: if faces is None:
return resp return resp
for face in faces: for face in faces:
# The detection output faces is a two-dimension array of type CV_32F, # pylint: disable=W0105
# whose rows are the detected face instances, """
# columns are the location of a face and 5 facial landmarks. The detection output faces is a two-dimension array of type CV_32F,
# The format of each row is as follows: whose rows are the detected face instances, columns are the location
# x1, y1, w, h, x_re, y_re, x_le, y_le, x_nt, y_nt, x_rcm, y_rcm, x_lcm, y_lcm, of a face and 5 facial landmarks.
# where x1, y1, w, h are the top-left coordinates, The format of each row is as follows:
# width and height of the face bounding box, x1, y1, w, h, x_re, y_re, x_le, y_le, x_nt, y_nt,
# {x, y}_{re, le, nt, rcm, lcm} stands for the coordinates of x_rcm, y_rcm, x_lcm, y_lcm,
# right eye, left eye, nose tip, the right corner and left corner where x1, y1, w, h are the top-left coordinates, width and height of
# of the mouth respectively. the face bounding box,
{x, y}_{re, le, nt, rcm, lcm} stands for the coordinates of right eye,
left eye, nose tip, the right corner and left corner of the mouth respectively.
"""
(x, y, w, h, x_re, y_re, x_le, y_le) = list(map(int, face[:8])) (x, y, w, h, x_re, y_re, x_le, y_le) = list(map(int, face[:8]))
# Yunet returns negative coordinates if it thinks part of
# the detected face is outside the frame.
# We set the coordinate to 0 if they are negative.
x = max(x, 0)
y = max(y, 0)
if resized: if resized:
image = original_image image = original_image
x, y, w, h = int(x / r), int(y / r), int(w / r), int(h / r) x, y, w, h = int(x / r), int(y / r), int(w / r), int(h / r)