Make gradient.shape==image.shape in the models.

fluid/adversarial/advbox/adversary.py

@@ -87,10 +87,12 @@ class Adversary(object):
 
         :return: bool
         """
+        assert adversarial_example is not None
+        assert self.__original.shape == adversarial_example.shape
+
         ok = self._is_successful(adversarial_label)
         if ok:
-            self.__adversarial_example = adversarial_example.reshape(
-                self.__original.shape)
+            self.__adversarial_example = adversarial_example
             self.__adversarial_label = adversarial_label
         return ok
 

fluid/adversarial/advbox/attacks/__init__.py

@@ -3,6 +3,7 @@ Attack methods
 """
 
 from .base import Attack
+from .deepfool import DeepFoolAttack
 from .gradientsign import FGSM
 from .gradientsign import GradientSignAttack
 from .iterator_gradientsign import IFGSM

fluid/adversarial/advbox/attacks/gradientsign.py

@@ -45,9 +45,8 @@ class GradientSignAttack(Attack):
                                              adversary.original_label)])
             gradient_sign = np.sign(gradient) * (max_ - min_)
 
-        original = adversary.original.reshape(gradient_sign.shape)
         for epsilon in epsilons:
-            adv_img = original + epsilon * gradient_sign
+            adv_img = adversary.original + epsilon * gradient_sign
             adv_img = np.clip(adv_img, min_, max_)
             adv_label = np.argmax(self.model.predict([(adv_img, 0)]))
             logging.info('epsilon = {:.3f}, pre_label = {}, adv_label={}'.

fluid/adversarial/advbox/attacks/iterator_gradientsign.py

@@ -35,7 +35,7 @@ class IteratorGradientSignAttack(Attack):
         min_, max_ = self.model.bounds()
 
         for epsilon in epsilons:
-            adv_img = None
+            adv_img = adversary.original
             for _ in range(steps):
                 if adversary.is_targeted_attack:
                     gradient = self.model.gradient([(adversary.original,
@@ -45,8 +45,6 @@ class IteratorGradientSignAttack(Attack):
                     gradient = self.model.gradient([(adversary.original,
                                                      adversary.original_label)])
                     gradient_sign = np.sign(gradient) * (max_ - min_)
-                if adv_img is None:
-                    adv_img = adversary.original.reshape(gradient_sign.shape)
                 adv_img = adv_img + gradient_sign * epsilon
                 adv_img = np.clip(adv_img, min_, max_)
                 adv_label = np.argmax(self.model.predict([(adv_img, 0)]))

fluid/adversarial/advbox/models/paddle.py

@@ -102,4 +102,4 @@ class PaddleModel(Model):
         grad, = self._exe.run(self._program,
                               feed=feeder.feed(image_batch),
                               fetch_list=[self._gradient])
-        return grad
+        return grad.reshape(image_batch[0][0].shape)