Harmonized opencv interface

bob/bio/face/embeddings/opencv.py

@@ -13,7 +13,7 @@ from bob.extension.download import get_file
 
 class OpenCVTransformer(TransformerMixin, BaseEstimator):
     """
-    Base Transformer using the OpenCV interface.
+    Base Transformer using the OpenCV DNN interface (https://docs.opencv.org/master/d2/d58/tutorial_table_of_content_dnn.html).
 
 
     .. note::
@@ -28,13 +28,24 @@ class OpenCVTransformer(TransformerMixin, BaseEstimator):
 
     config:
         Path containing some configuration file (e.g. .json, .prototxt)
+
+    preprocessor:
+        A function that will transform the data right before forward. The default transformation is `X/255`
+
     """
 
-    def __init__(self, checkpoint_path=None, config=None, **kwargs):
+    def __init__(
+        self,
+        checkpoint_path=None,
+        config=None,
+        preprocessor=lambda x: x / 255,
+        **kwargs,
+    ):
         super().__init__(**kwargs)
         self.checkpoint_path = checkpoint_path
         self.config = config
         self.model = None
+        self.preprocessor = preprocessor
 
     def _load_model(self):
         import cv2
@@ -63,14 +74,11 @@ class OpenCVTransformer(TransformerMixin, BaseEstimator):
         if self.model is None:
             self._load_model()
 
-        import ipdb
-
-        ipdb.set_trace()
+        X = check_array(X, allow_nd=True)
 
-        img = np.array(X)
-        img = img / 255
+        X = self.preprocessor(X)
 
-        self.model.setInput(img)
+        self.model.setInput(X)
 
         return self.model.forward()
 
@@ -108,10 +116,30 @@ class VGG16_Oxford(OpenCVTransformer):
         config = os.path.join(path, "vgg_face_caffe", "VGG_FACE_deploy.prototxt")
         checkpoint_path = os.path.join(path, "vgg_face_caffe", "VGG_FACE.caffemodel")
 
-        super(VGG16_Oxford, self).__init__(checkpoint_path, config)
+        caffe_average_img = [129.1863, 104.7624, 93.5940]
+
+        def preprocessor(X):
+            """
+            Normalize using data from caffe
+
+            Caffe has the shape `C x H x W` and the chanel is BGR and 
+
+            """
+            # To BGR
+            X = X[:, ::-1, :, :].astype("float32")
+
+            # Subtracting
+            X[:, :, :, 0] -= caffe_average_img[0]
+            X[:, :, :, 1] -= caffe_average_img[1]
+            X[:, :, :, 2] -= caffe_average_img[2]
+
+            return X
+
+        super(VGG16_Oxford, self).__init__(checkpoint_path, config, preprocessor)
 
     def _load_model(self):
         import cv2
 
         net = cv2.dnn.readNet(self.checkpoint_path, self.config)
         self.model = net
+