Rethinking baselines

bob/bio/face/config/transformers/eyes_crop/facenet.py

+import bob.bio.face
+
+from sklearn.pipeline import make_pipeline
+from bob.bio.base.wrappers import wrap_sample_preprocessor
+from bob.pipelines import wrap
+from bob.bio.face.transformers import FaceNetSanderberg
+
+
+# This is the size of the image that this model expects
+CROPPED_IMAGE_HEIGHT = 160
+CROPPED_IMAGE_WIDTH = 160
+
+
+# eye positions for frontal images
+RIGHT_EYE_POS = (46, 53)
+LEFT_EYE_POS = (46, 107)
+
+
+legacy_face_cropper = bob.bio.face.preprocessor.FaceCrop(
+    cropped_image_size=(CROPPED_IMAGE_HEIGHT, CROPPED_IMAGE_WIDTH),
+    cropped_positions={"leye": LEFT_EYE_POS, "reye": RIGHT_EYE_POS},
+    color_channel="rgb",
+)
+
+embedding = FaceNetSanderberg()
+
+transformer = make_pipeline(
+    wrap_sample_preprocessor(
+        legacy_face_cropper,
+        transform_extra_arguments=(("annotations", "annotations"),),
+    ),
+    wrap(["sample"], embedding),
+)

bob/bio/face/test/test_pipelines.py

+from bob.extension.config import load
+import pkg_resources
+import numpy as np
+from bob.pipelines import Sample, SampleSet
+
+
+def get_fake_sample(face_size=(160, 160), eyes={"leye": (46, 107), "reye": (46, 53)}):
+
+    data = np.random.rand(3,400,400)
+    annotations={"leye": (115, 267), "reye": (115, 132)}
+    return Sample(data, key="1", annotations=annotations)
+
+
+def test_facenet_pipeline():
+    config_name = pkg_resources.resource_filename('bob.bio.face', 'config/transformers/eyes_crop/facenet.py')
+    transformer = load([config_name]).transformer
+
+    import ipdb; ipdb.set_trace()
+    fake_sample = get_fake_sample()
+    transformed_sample = transformer.transform([fake_sample])[0]
+    assert transformed_sample.data.size == 160
+
+    pass
\ No newline at end of file

bob/bio/face/transformers/facenet_sanderberg.py

@@ -165,17 +165,25 @@ class FaceNetSanderberg(TransformerMixin, BaseEstimator):
         logger.info("Successfully loaded the model.")
         self.loaded = True
 
-    def transform(self, img):
-        images = self._check_feature(img)
-        if not self.loaded:
-            self.load_model()
-
-        feed_dict = {
-            self.images_placeholder: images,
-            self.phase_train_placeholder: False,
-        }
-        features = self.session.run(self.embeddings, feed_dict=feed_dict)
-        return features.flatten()
+    def transform(self, X):
+        def _transform(X):
+        
+            import ipdb; ipdb.set_trace()
+            images = self._check_feature(X)
+            if not self.loaded:
+                self.load_model()
+
+            feed_dict = {
+                self.images_placeholder: images,
+                self.phase_train_placeholder: False,
+            }
+            features = self.session.run(self.embeddings, feed_dict=feed_dict)
+            return features.flatten()
+
+        if isinstance(X, list):
+            return [_transform(i) for i in X]
+        else:
+            return _transform(X)
 
     @staticmethod
     def get_modelpath():
@@ -214,3 +222,9 @@ class FaceNetSanderberg(TransformerMixin, BaseEstimator):
         d.pop("phase_train_placeholder") if "phase_train_placeholder" in d else None
         tf.compat.v1.reset_default_graph()
         return d
+
+    def _more_tags(self):
+        return {"stateless": True, "requires_fit": False}
+
+    def fit(self, X, y=None):
+        return self

bob/bio/face/transformers/tensorflow_compat_v1.py

@@ -160,3 +160,11 @@ class TensorflowCompatV1(TransformerMixin, BaseEstimator):
             bob.io.base.create_directories_safe(model_path)
             zip_file = os.path.join(model_path, zip_file)
             bob.extension.download.download_and_unzip(urls, zip_file)
+
+
+    def fit(self, X, y=None):
+        return self
+
+    def _more_tags(self):
+        return {"stateless": True, "requires_fit": False}
+