Added MixUp augmentation

albumentations/augmentations/geometric/transforms.py

@@ -34,6 +34,7 @@
     "OpticalDistortion",
     "GridDistortion",
     "PadIfNeeded",
+    "MixUp"
 ]
 
 
@@ -1491,3 +1492,55 @@ def get_params_dependent_on_targets(self, params):
 
     def get_transform_init_args_names(self):
         return "num_steps", "distort_limit", "interpolation", "border_mode", "value", "mask_value", "normalized"
+
+
+class MixUp(DualTransform):
+    """Generates a weighted combination of the pair of input images.
+    The bboxes are kept as in the input images
+
+    Args:
+        alpha (float): Alpha for distribution
+        beta (float): Beta for distribution
+    Targets:
+        image, image1, bboxes, bboxes1
+    Image types:
+        uint8, float32
+    Reference:
+        Hongyi Zhang, Moustapha Cisse, Yann N. Dauphin, David Lopez-Paz:
+        "mixup: Beyond Empirical Risk Minimization"
+    """
+
+    def __init__(
+        self,
+        alpha = 32.,
+        beta = 32.,
+        always_apply: bool = False,
+        p: float = 1,
+    ):
+        super(MixUp, self).__init__(always_apply=always_apply, p=p)
+        self.alpha = alpha
+        self.beta = beta
+
+    def apply(self, image: np.ndarray, image1: np.ndarray, r, **params) -> np.ndarray:
+        h1, w1, _ = image.shape
+        h2, w2, _ = image1.shape
+        if h1 != h2 or w1 != w2:
+            raise TypeError("MixUp transformation expects both images to have identical shape.")
+        im = (image * r + image1 * (1 - r))
+        return im
+
+    def get_params(self):
+        return {"r": np.random.beta(self.alpha, self.beta)}  # draw samples from a Beta distribution.
+
+    def apply_to_bboxes(self, bboxes: Sequence[BoxType], bboxes1: Sequence[BoxType], **params) -> List[BoxType]:
+        return bboxes + bboxes1
+
+    def get_params_dependent_on_targets(self, params: Dict[str, Any]) -> Dict[str, Any]:
+        return {
+            "image1": params['image1'], 
+            "bboxes1": params['bboxes1']
+        }
+
+    @property
+    def targets_as_params(self) -> List[str]:
+        return ["image", "image1", "bboxes", "bboxes1"]