python3 support

advbox/attacks/cw.py

@@ -316,7 +316,7 @@ def _predict_adv(self, img_constrained):
         adv_logits = self.model.predict(img_constrained)
         adv_label = np.argmax(adv_logits)
 
-        return adv_label, adv_logits[0][adv_label] # adv_lab, adv_score
+        return adv_label, adv_logits[adv_label] # adv_lab, adv_score
 
 
 CW_L2 = CW_L2_Attack

example/imagenet_example_fgsm.py

@@ -23,7 +23,7 @@
 import logging
 import numpy as np
 import paddle.fluid as fluid
-import paddle.v2 as paddle
+import paddle
 
 #classification
 import models

example/models/alexnet.py

@@ -1,3 +1,4 @@
+from builtins import object
 import paddle
 import paddle.fluid as fluid
 import math
@@ -18,7 +19,7 @@
 }
 
 
-class AlexNet():
+class AlexNet(object):
     def __init__(self):
         self.params = train_parameters
 

example/models/resnet.py

@@ -1,3 +1,8 @@
+from __future__ import division
+from builtins import str
+from builtins import range
+from builtins import object
+from past.utils import old_div
 import paddle
 import paddle.fluid as fluid
 import math
@@ -58,7 +63,7 @@ def create_parameter(layers, shape, dtype):
 
     return scale, bias, mean, variance
 
-class ResNet():
+class ResNet(object):
     def __init__(self, layers=50):
         self.params = train_parameters
         self.layers = layers
@@ -129,7 +134,7 @@ def conv_bn_layer(self,
             num_filters=num_filters,
             filter_size=filter_size,
             stride=stride,
-            padding=int((filter_size - 1) / 2),
+            padding=int(old_div((filter_size - 1), 2)),
             groups=groups,
             param_attr=fluid.ParamAttr(name=param_name + '.w' + '_0'),
             act=None,

example/reader.py

@@ -1,3 +1,4 @@
+from __future__ import division
 # Copyright 2017 - 2018 Baidu Inc.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -12,6 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from past.utils import old_div
 import os
 import random
 import functools
@@ -40,8 +42,8 @@ def crop_image(img, target_size, center):
     width, height = img.size
     size = target_size
     if center == True:
-        w_start = (width - size) / 2
-        h_start = (height - size) / 2
+        w_start = old_div((width - size), 2)
+        h_start = old_div((height - size), 2)
     else:
         w_start = random.randint(0, width - size)
         h_start = random.randint(0, height - size)
@@ -62,7 +64,7 @@ def process_image(sample):
     if img.mode != 'RGB':
         img = img.convert('RGB')
 
-    img = np.array(img).astype('float32').transpose((2, 0, 1)) / 255
+    img = old_div(np.array(img).astype('float32').transpose((2, 0, 1)), 255)
     img -= img_mean
     img /= img_std
     return [img], img_path