alexnet network update.

model_zoo/official/cv/alexnet/eval.py

@@ -50,7 +50,7 @@ if __name__ == "__main__":
 
     if args.dataset_name == 'cifar10':
         cfg = alexnet_cifar10_cfg
-        network = AlexNet(cfg.num_classes)
+        network = AlexNet(cfg.num_classes, phase='test')
         loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
         opt = nn.Momentum(network.trainable_params(), cfg.learning_rate, cfg.momentum)
         ds_eval = create_dataset_cifar10(args.data_path, cfg.batch_size, status="test", target=args.device_target)
@@ -64,7 +64,7 @@ if __name__ == "__main__":
 
     elif args.dataset_name == 'imagenet':
         cfg = alexnet_imagenet_cfg
-        network = AlexNet(cfg.num_classes)
+        network = AlexNet(cfg.num_classes, phase='test')
         loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
         ds_eval = create_dataset_imagenet(args.data_path, cfg.batch_size, training=False)
 

model_zoo/official/cv/alexnet/src/alexnet.py

@@ -14,43 +14,38 @@
 # ============================================================================
 """Alexnet."""
 import mindspore.nn as nn
-from mindspore.common.initializer import TruncatedNormal
 from mindspore.ops import operations as P
 
-def conv(in_channels, out_channels, kernel_size, stride=1, padding=0, pad_mode="valid"):
-    weight = weight_variable()
-    return nn.Conv2d(in_channels, out_channels,
-                     kernel_size=kernel_size, stride=stride, padding=padding,
-                     weight_init=weight, has_bias=False, pad_mode=pad_mode)
-
-def fc_with_initialize(input_channels, out_channels):
-    weight = weight_variable()
-    bias = weight_variable()
-    return nn.Dense(input_channels, out_channels, weight, bias)
-
-def weight_variable():
-    return TruncatedNormal(0.02)
+def conv(in_channels, out_channels, kernel_size, stride=1, padding=0, pad_mode="valid", has_bias=True):
+    return nn.Conv2d(in_channels, out_channels, kernel_size=kernel_size, stride=stride, padding=padding,
+                     has_bias=has_bias, pad_mode=pad_mode)
 
+def fc_with_initialize(input_channels, out_channels, has_bias=True):
+    return nn.Dense(input_channels, out_channels, has_bias=has_bias)
 
 class AlexNet(nn.Cell):
     """
     Alexnet
     """
-    def __init__(self, num_classes=10, channel=3, include_top=True):
+    def __init__(self, num_classes=10, channel=3, phase='train', include_top=True):
         super(AlexNet, self).__init__()
-        self.conv1 = conv(channel, 96, 11, stride=4)
-        self.conv2 = conv(96, 256, 5, pad_mode="same")
-        self.conv3 = conv(256, 384, 3, pad_mode="same")
-        self.conv4 = conv(384, 384, 3, pad_mode="same")
-        self.conv5 = conv(384, 256, 3, pad_mode="same")
-        self.relu = nn.ReLU()
-        self.max_pool2d = P.MaxPool(ksize=3, strides=2)
+        self.conv1 = conv(channel, 64, 11, stride=4, pad_mode="same", has_bias=True)
+        self.conv2 = conv(64, 192, 5, pad_mode="same", has_bias=True)
+        self.conv3 = conv(192, 384, 3, pad_mode="same", has_bias=True)
+        self.conv4 = conv(384, 256, 3, pad_mode="same", has_bias=True)
+        self.conv5 = conv(256, 256, 3, pad_mode="same", has_bias=True)
+        self.relu = P.ReLU()
+        self.max_pool2d = nn.MaxPool2d(kernel_size=3, stride=2, pad_mode='valid')
         self.include_top = include_top
         if self.include_top:
+            dropout_ratio = 0.65
+            if phase == 'test':
+                dropout_ratio = 1.0
             self.flatten = nn.Flatten()
             self.fc1 = fc_with_initialize(6 * 6 * 256, 4096)
             self.fc2 = fc_with_initialize(4096, 4096)
             self.fc3 = fc_with_initialize(4096, num_classes)
+            self.dropout = nn.Dropout(dropout_ratio)
 
     def construct(self, x):
         """define network"""
@@ -72,7 +67,9 @@ class AlexNet(nn.Cell):
         x = self.flatten(x)
         x = self.fc1(x)
         x = self.relu(x)
+        x = self.dropout(x)
         x = self.fc2(x)
         x = self.relu(x)
+        x = self.dropout(x)
         x = self.fc3(x)
         return x

model_zoo/official/cv/alexnet/src/config.py

@@ -38,8 +38,8 @@ alexnet_imagenet_cfg = edict({
     'epoch_size': 150,
     'batch_size': 256,
     'buffer_size': None, # invalid parameter
-    'image_height': 227,
-    'image_width': 227,
+    'image_height': 224,
+    'image_width': 224,
     'save_checkpoint_steps': 625,
     'keep_checkpoint_max': 10,
 

model_zoo/official/cv/alexnet/train.py

@@ -94,7 +94,7 @@ if __name__ == "__main__":
     if ds_train.get_dataset_size() == 0:
         raise ValueError("Please check dataset size > 0 and batch_size <= dataset size")
 
-    network = AlexNet(cfg.num_classes)
+    network = AlexNet(cfg.num_classes, phase='train')
 
     loss_scale_manager = None
     metrics = None