using type hinting for backend functions

deepface/DeepFace.py

@@ -889,8 +889,12 @@ def extract_faces(
 
 @deprecated(version="0.0.78", reason="Use DeepFace.extract_faces instead of DeepFace.detectFace")
 def detectFace(
-    img_path, target_size=(224, 224), detector_backend="opencv", enforce_detection=True, align=True
-):
+    img_path: Union[str, np.ndarray],
+    target_size: tuple = (224, 224),
+    detector_backend: str = "opencv",
+    enforce_detection: bool = True,
+    align: bool = True,
+) -> np.ndarray:
     """
     Deprecated function. Use extract_faces for same functionality.
 
@@ -942,7 +946,7 @@ def detectFace(
 functions.initialize_folder()
 
 
-def cli():
+def cli() -> None:
     """
     command line interface function will be offered in this block
     """

deepface/basemodels/ArcFace.py

@@ -14,8 +14,8 @@
 tf_version = int(tf.__version__.split(".", maxsplit=1)[0])
 
 if tf_version == 1:
+    from keras.models import Model
     from keras.engine import training
-    import keras
     from keras.layers import (
         ZeroPadding2D,
         Input,
@@ -28,8 +28,8 @@
         Dense,
     )
 else:
+    from tensorflow.keras.models import Model
     from tensorflow.python.keras.engine import training
-    from tensorflow import keras
     from tensorflow.keras.layers import (
         ZeroPadding2D,
         Input,
@@ -41,15 +41,11 @@
         Flatten,
         Dense,
     )
-# --------------------------------
-
-
-# url = "https://drive.google.com/uc?id=1LVB3CdVejpmGHM28BpqqkbZP5hDEcdZY"
 
 
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/arcface_weights.h5",
-):
+) -> Model:
     base_model = ResNet34()
     inputs = base_model.inputs[0]
     arcface_model = base_model.outputs[0]
@@ -62,7 +58,7 @@ def loadModel(
     embedding = BatchNormalization(momentum=0.9, epsilon=2e-5, name="embedding", scale=True)(
         arcface_model
     )
-    model = keras.models.Model(inputs, embedding, name=base_model.name)
+    model = Model(inputs, embedding, name=base_model.name)
 
     # ---------------------------------------
     # check the availability of pre-trained weights
@@ -84,7 +80,7 @@ def loadModel(
     return model
 
 
-def ResNet34():
+def ResNet34() -> Model:
 
     img_input = Input(shape=(112, 112, 3))
 

deepface/basemodels/DeepID.py

@@ -41,7 +41,7 @@
 
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/deepid_keras_weights.h5",
-):
+) -> Model:
 
     myInput = Input(shape=(55, 47, 3))
 

deepface/basemodels/DlibResNet.py

@@ -13,8 +13,14 @@
 class DlibResNet:
     def __init__(self):
 
-        # this is not a must dependency
-        import dlib  # 19.20.0
+        ## this is not a must dependency. do not import it in the global level.
+        try:
+            import dlib
+        except ModuleNotFoundError as e:
+            raise ImportError(
+                "Dlib is an optional dependency, ensure the library is installed."
+                "Please install using 'pip install dlib' "
+            ) from e
 
         self.layers = [DlibMetaData()]
 
@@ -49,7 +55,7 @@ def __init__(self):
 
         # return None  # classes must return None
 
-    def predict(self, img_aligned):
+    def predict(self, img_aligned: np.ndarray) -> np.ndarray:
 
         # functions.detectFace returns 4 dimensional images
         if len(img_aligned.shape) == 4:

deepface/basemodels/DlibWrapper.py

@@ -1,5 +1,6 @@
+from typing import Any
 from deepface.basemodels.DlibResNet import DlibResNet
 
 
-def loadModel():
+def loadModel() -> Any:
     return DlibResNet()

deepface/basemodels/Facenet.py

@@ -47,7 +47,7 @@ def scaling(x, scale):
     return x * scale
 
 
-def InceptionResNetV2(dimension=128):
+def InceptionResNetV2(dimension=128) -> Model:
 
     inputs = Input(shape=(160, 160, 3))
     x = Conv2D(32, 3, strides=2, padding="valid", use_bias=False, name="Conv2d_1a_3x3")(inputs)
@@ -1618,12 +1618,9 @@ def InceptionResNetV2(dimension=128):
     return model
 
 
-# url = 'https://drive.google.com/uc?id=1971Xk5RwedbudGgTIrGAL4F7Aifu7id1'
-
-
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/facenet_weights.h5",
-):
+) -> Model:
     model = InceptionResNetV2()
 
     # -----------------------------------

deepface/basemodels/Facenet512.py

@@ -1,14 +1,23 @@
 import os
 import gdown
+import tensorflow as tf
 from deepface.basemodels import Facenet
 from deepface.commons import functions
 from deepface.commons.logger import Logger
 
 logger = Logger(module="basemodels.Facenet512")
 
+tf_version = int(tf.__version__.split(".", maxsplit=1)[0])
+
+if tf_version == 1:
+    from keras.models import Model
+else:
+    from tensorflow.keras.models import Model
+
+
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/facenet512_weights.h5",
-):
+) -> Model:
 
     model = Facenet.InceptionResNetV2(dimension=512)
 

deepface/basemodels/FbDeepFace.py

@@ -40,7 +40,7 @@
 
 def loadModel(
     url="https://github.com/swghosh/DeepFace/releases/download/weights-vggface2-2d-aligned/VGGFace2_DeepFace_weights_val-0.9034.h5.zip",
-):
+) -> Model:
     base_model = Sequential()
     base_model.add(
         Convolution2D(32, (11, 11), activation="relu", name="C1", input_shape=(152, 152, 3))

deepface/basemodels/OpenFace.py

@@ -27,7 +27,7 @@
 
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/openface_weights.h5",
-):
+) -> Model:
     myInput = Input(shape=(96, 96, 3))
 
     x = ZeroPadding2D(padding=(3, 3), input_shape=(96, 96, 3))(myInput)

deepface/basemodels/SFace.py

@@ -1,4 +1,6 @@
 import os
+from typing import Any
+
 import numpy as np
 import cv2 as cv
 import gdown
@@ -25,7 +27,7 @@ def __init__(self, model_path):
 
         self.layers = [_Layer()]
 
-    def predict(self, image):
+    def predict(self, image: np.ndarray) -> np.ndarray:
         # Preprocess
         input_blob = (image[0] * 255).astype(
             np.uint8
@@ -39,7 +41,7 @@ def predict(self, image):
 
 def load_model(
     url="https://github.com/opencv/opencv_zoo/raw/main/models/face_recognition_sface/face_recognition_sface_2021dec.onnx",
-):
+) -> Any:
 
     home = functions.get_deepface_home()
 

deepface/basemodels/VGGFace.py

@@ -34,7 +34,7 @@
 # ---------------------------------------
 
 
-def baseModel():
+def baseModel() -> Sequential:
     model = Sequential()
     model.add(ZeroPadding2D((1, 1), input_shape=(224, 224, 3)))
     model.add(Convolution2D(64, (3, 3), activation="relu"))
@@ -83,31 +83,21 @@ def baseModel():
     return model
 
 
-# url = 'https://drive.google.com/uc?id=1CPSeum3HpopfomUEK1gybeuIVoeJT_Eo'
-
-
 def loadModel(
     url="https://github.com/serengil/deepface_models/releases/download/v1.0/vgg_face_weights.h5",
-):
+) -> Model:
 
     model = baseModel()
 
-    # -----------------------------------
-
     home = functions.get_deepface_home()
     output = home + "/.deepface/weights/vgg_face_weights.h5"
 
     if os.path.isfile(output) != True:
         logger.info("vgg_face_weights.h5 will be downloaded...")
         gdown.download(url, output, quiet=False)
 
-    # -----------------------------------
-
     model.load_weights(output)
 
-    # -----------------------------------
-
-    # TO-DO: why?
     vgg_face_descriptor = Model(inputs=model.layers[0].input, outputs=model.layers[-2].output)
 
     return vgg_face_descriptor

deepface/commons/distance.py

@@ -1,14 +1,25 @@
+from typing import Union
 import numpy as np
 
 
-def findCosineDistance(source_representation, test_representation):
+def findCosineDistance(
+    source_representation: Union[np.ndarray, list], test_representation: Union[np.ndarray, list]
+) -> np.float64:
+    if isinstance(source_representation, list):
+        source_representation = np.array(source_representation)
+
+    if isinstance(test_representation, list):
+        test_representation = np.array(test_representation)
+
     a = np.matmul(np.transpose(source_representation), test_representation)
     b = np.sum(np.multiply(source_representation, source_representation))
     c = np.sum(np.multiply(test_representation, test_representation))
     return 1 - (a / (np.sqrt(b) * np.sqrt(c)))
 
 
-def findEuclideanDistance(source_representation, test_representation):
+def findEuclideanDistance(
+    source_representation: Union[np.ndarray, list], test_representation: Union[np.ndarray, list]
+) -> np.float64:
     if isinstance(source_representation, list):
         source_representation = np.array(source_representation)
 
@@ -21,11 +32,11 @@ def findEuclideanDistance(source_representation, test_representation):
     return euclidean_distance
 
 
-def l2_normalize(x):
+def l2_normalize(x: np.ndarray) -> np.ndarray:
     return x / np.sqrt(np.sum(np.multiply(x, x)))
 
 
-def findThreshold(model_name, distance_metric):
+def findThreshold(model_name: str, distance_metric: str) -> float:
 
     base_threshold = {"cosine": 0.40, "euclidean": 0.55, "euclidean_l2": 0.75}