update

Author: foamliu

densenet121.py

@@ -1,9 +1,8 @@
 # -*- coding: utf-8 -*-
 
 from keras.optimizers import SGD
-from keras.layers import Input, merge, ZeroPadding2D
+from keras.layers import Input, merge, ZeroPadding2D, Conv2D
 from keras.layers.core import Dense, Dropout, Activation
-from keras.layers.convolutional import Convolution2D
 from keras.layers.pooling import AveragePooling2D, GlobalAveragePooling2D, MaxPooling2D
 from keras.layers.normalization import BatchNormalization
 from keras.models import Model
@@ -57,7 +56,7 @@ def densenet121_model(img_rows, img_cols, color_type=1, nb_dense_block=4, growth
 
     # Initial convolution
     x = ZeroPadding2D((3, 3), name='conv1_zeropadding')(img_input)
-    x = Convolution2D(nb_filter, 7, 7, subsample=(2, 2), name='conv1', bias=False)(x)
+    x = Conv2D(nb_filter, (7, 7), strides=(2, 2), name='conv1', use_bias=False)(x)
     x = BatchNormalization(epsilon=eps, axis=concat_axis, name='conv1_bn')(x)
     x = Scale(axis=concat_axis, name='conv1_scale')(x)
     x = Activation('relu', name='relu1')(x)
@@ -88,10 +87,10 @@ def densenet121_model(img_rows, img_cols, color_type=1, nb_dense_block=4, growth
 
     if K.image_dim_ordering() == 'th':
       # Use pre-trained weights for Theano backend
-      weights_path = 'imagenet_models/densenet121_weights_th.h5'
+      weights_path = 'models/densenet121_weights_th.h5'
     else:
       # Use pre-trained weights for Tensorflow backend
-      weights_path = 'imagenet_models/densenet121_weights_tf.h5'
+      weights_path = 'models/densenet121_weights_tf.h5'
 
     model.load_weights(weights_path, by_name=True)
 
@@ -130,7 +129,7 @@ def conv_block(x, stage, branch, nb_filter, dropout_rate=None, weight_decay=1e-4
     x = BatchNormalization(epsilon=eps, axis=concat_axis, name=conv_name_base+'_x1_bn')(x)
     x = Scale(axis=concat_axis, name=conv_name_base+'_x1_scale')(x)
     x = Activation('relu', name=relu_name_base+'_x1')(x)
-    x = Convolution2D(inter_channel, 1, 1, name=conv_name_base+'_x1', bias=False)(x)
+    x = Conv2D(inter_channel, (1, 1), name=conv_name_base+'_x1', use_bias=False)(x)
 
     if dropout_rate:
         x = Dropout(dropout_rate)(x)
@@ -140,7 +139,7 @@ def conv_block(x, stage, branch, nb_filter, dropout_rate=None, weight_decay=1e-4
     x = Scale(axis=concat_axis, name=conv_name_base+'_x2_scale')(x)
     x = Activation('relu', name=relu_name_base+'_x2')(x)
     x = ZeroPadding2D((1, 1), name=conv_name_base+'_x2_zeropadding')(x)
-    x = Convolution2D(nb_filter, 3, 3, name=conv_name_base+'_x2', bias=False)(x)
+    x = Conv2D(nb_filter, (3, 3), name=conv_name_base+'_x2', use_bias=False)(x)
 
     if dropout_rate:
         x = Dropout(dropout_rate)(x)
@@ -167,7 +166,7 @@ def transition_block(x, stage, nb_filter, compression=1.0, dropout_rate=None, we
     x = BatchNormalization(epsilon=eps, axis=concat_axis, name=conv_name_base+'_bn')(x)
     x = Scale(axis=concat_axis, name=conv_name_base+'_scale')(x)
     x = Activation('relu', name=relu_name_base)(x)
-    x = Convolution2D(int(nb_filter * compression), 1, 1, name=conv_name_base, bias=False)(x)
+    x = Conv2D(int(nb_filter * compression), (1, 1), name=conv_name_base, use_bias=False)(x)
 
     if dropout_rate:
         x = Dropout(dropout_rate)(x)