BNSTechnologyDevelopment
/
Jiale
6
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

change readmeW.md to readme.md

This commit is contained in:
DiaosWang 2024-05-22 17:16:30 +08:00
parent 93c597dd94
commit b1a4a892b4
3 changed files with 121 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
proj_deepface/readme.md Normal file
View File

@ -0,0 +1,54 @@
ref: https://github.com/serengil/deepface
# model store
model checkpoint 放在了 `C:\Users\User\.deepface\weights`
# installation
```
$ pip install deepface
```
# Usage:
```
(deepface_v2) PS path/to/workspace> python
Python 3.7.16 (default, Jan 17 2023, 16:06:28) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> from deepface import DeepFace
```
# Model introduce
```
Deepface is a hybrid face recognition package. It currently wraps many state-of-the-art face recognition models:
model_name Declared LFW Score
VGG-Face 98.9%
Facenet 99.2%
Facenet512 99.6% drop
OpenFace 92.9%
DeepID 97.4%
Dlib 99.3%
SFace 99.5%
ArcFace 99.5% 阈值调整为1.00 **
GhostFaceNet 99.7% **
Human-beings 97.5%
The default configuration uses VGG-Face model.
```
```
backends = {
"opencv": OpenCv.OpenCvClient,
"mtcnn": MtCnn.MtCnnClient,
"ssd": Ssd.SsdClient,
"dlib": Dlib.DlibClient,
"retinaface": RetinaFace.RetinaFaceClient,
"mediapipe": MediaPipe.MediaPipeClient,
"yolov8": Yolo.YoloClient,
"yunet": YuNet.YuNetClient,
"fastmtcnn": FastMtCnn.FastMtCnnClient,
}
```
```
distance_metric = ['euclidean_l2','cosine','euclidean']
author: Euclidean L2 form seems to be more stable than cosine and regular Euclidean distance based on experiments.
```

14
proj_deepface/readmeW.md
View File

@ -1,14 +0,0 @@
ref: https://github.com/serengil/deepface
# store
model checkpoint 放在了 `C:\Users\JIALE\.deepface\weights`
# installation
# Usage:
(deepface_v2) PS C:\Users\JIALE\Desktop\bns_proj\proj_deepface> python
Python 3.7.16 (default, Jan 17 2023, 16:06:28) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> from deepface import DeepFace

103
proj_deepface/webmain.py
View File

@ -51,6 +51,7 @@ class Status(Enum):
BASE64_EMPTY = "base64 is empty"
NotImplementedError = "Not Implemented Error"
ADD_HAS_EXISTED = "when adding, file has existed"
DB_NO_FACE = "face_db has no face"
GLOBAL_DICT = {
"ADD": Opt.ADD,
@ -65,6 +66,7 @@ GLOBAL_DICT = {
}
class FaceHelper:
def __init__(self, db_path: str, detector_backend: str, distance_metric: str, model_name: str):
self.db_path = "C:/Users/JIALE/Desktop/bns_proj/proj_deepface/face"
self.detector_backend = "opencv"
@ -74,7 +76,11 @@ class FaceHelper:
self.target_size = self.model.input_shape
self.threshold = verification.find_threshold(model_name, distance_metric)
self.pkl_name = f"ds_{self.model_name}_{self.detector_backend}_v2.pkl".replace("-", "").lower()
self.dirty = True
self.db_representations = None
self.keepDBLive()
def get_facial_areas(self, img: np.ndarray):
"""
Find facial area coordinates in the given image
@ -88,8 +94,6 @@ class FaceHelper:
detected_face
descriptions
"""
# print("get facial areas, img.shape: ", img.shape)
# beg = time.time()
try:
face_objs = detection.extract_faces(
img_path = img,
@ -202,8 +206,7 @@ class FaceHelper:
if retjson['status'] == 0:
return retjson
# db_imgs = os.listdir(self.db_path)
db_imgs = [db_img.split(".")[0] for db_img in os.listdir(self.db_path) if db_img.endswith(".jpg")]
if opt_mode == Opt.ADD: # add
if img_name in db_imgs:
@ -215,35 +218,19 @@ class FaceHelper:
return Status.NotImplementedError
# update the face_db to keep latest
search_identity( # 保证database是最新的
detected_face=np.zeros([224, 224, 3]),
db_path=self.db_path,
detector_backend=self.detector_backend,
distance_metric=self.distance_metric,
model_name=self.model_name,
)
self.dirty = True
self.keepDBLive()
return retjson
#interface 2: face recognition using embeddings on face_db
def faceRecog(self, img: Union[str,np.ndarray], img_mode: Ftype):
# keep the database is the newest
search_identity(
detected_face=np.zeros([224, 224, 3]),
db_path=self.db_path,
detector_backend=self.detector_backend,
distance_metric=self.distance_metric,
model_name=self.model_name,
)
if self.dirty:
self.keepDBLive()
if len(self.db_representations) <= 0:
return Status.DB_NO_FACE
# open and read face db
datastore_path = os.path.join(self.db_path, self.pkl_name)
with open(datastore_path, "rb") as f:
db_representations = pickle.load(f)
assert len(db_representations) > 0, 'no face in database'
print(f"{len(db_representations)} faces representataion in {self.pkl_name}.")
# process the img
ret = FaceHelper.process_img(img, img_mode)
if isinstance(ret, Status):
@ -251,7 +238,7 @@ class FaceHelper:
target_img = ret
# begin face recognition
df = pd.DataFrame(db_representations)
df = pd.DataFrame(self.db_representations)
labels = []
face_coors, detected_faces, descriptions = self.get_facial_areas(target_img)
@ -292,8 +279,27 @@ class FaceHelper:
success, encoded_target_img = cv2.imencode('.png', target_img)
if success:
target_img_base64 = base64.b64encode(encoded_target_img).decode("utf-8")
return labels, face_coors, target_img_base64
def keepDBLive(self):
# keep the database is the newest
search_identity(
detected_face=np.zeros([224, 224, 3]),
db_path=self.db_path,
detector_backend=self.detector_backend,
distance_metric=self.distance_metric,
model_name=self.model_name,
)
self.dirty = False
# load the representations from face_db
datestore_path = os.path.join(self.db_path, self.pkl_name)
with open(datestore_path, "rb") as f:
self.db_representations = pickle.load(f)
return len(self.db_representations)
facehelper = FaceHelper(
db_path = "C:/Users/JIALE/Desktop/bns_proj/proj_deepface/face",
@ -305,22 +311,47 @@ facehelper = FaceHelper(
app = FastAPI()
class item(BaseModel):
type: str
TYPE: str
img: str
img_name: str
class config(BaseModel):
db_path: str # default: "C:/Users/JIALE/Desktop/bns_proj/proj_deepface/face"
detector_backend: str #default:
distance_metric: str
model_name: str
# @app.post("/master/")
# def master(input: item ):
# return {"type": item["type"], "img":"fdasfsdf"}
@app.post("/init/")
def init(input: config):
facehelper = FaceHelper(
db_path = input.db_path,
detector_backend = input.detector_backend,
distance_metric = input.distance_metric,
model_name = input.model_name,
)
@app.get("/keepDBLastest")
def keepDBLastest():
ret = facehelper.keepDBLive()
return {"status":"the faceDB is lastest", "description": f"db has {ret} faces"}
@app.post("/interface/")
def main(input: item):
interface_type, img_mode, opt_mode = input.type.split("_")
img_mode = GLOBAL_DICT[img_mode]
if interface_type == "UPDATEDB":
vars = input.TYPE.split("_")
if len(vars) != 3:
return {"status": 0, "description":"wrong TYPE"}
interface_type, img_mode, opt_mode = vars
try:
img_mode = GLOBAL_DICT[img_mode]
except KeyError:
return {"status": 0, "description": "wrong img_mode"}
try:
opt_mode = GLOBAL_DICT[opt_mode]
except KeyError:
return {"status": 0, "description": "wrong opt_mode"}
if interface_type == "UPDATEDB":
# call updateFaceDB for handling the picture
ret = facehelper.updateFaceDB(img = input.img,
img_name = input.img_name,