tensorflow - 找到卷积层和密集层的数量
问题描述
我从 Kaggle 复制代码,我无法计算其中的层数。我正在研究图像分类模型。谁能解释一下。我尝试了大多数解决方案,但我无法计算卷积层和密集层。
model = Sequential()
inputShape = (height, width, depth)
chanDim = -1
if K.image_data_format() == "channels_first":
inputShape = (depth, height, width)
chanDim = 1
model.add(Conv2D(32, (3, 3), padding="same",input_shape=inputShape))
model.add(Activation("relu"))
model.add(BatchNormalization(axis=chanDim))
model.add(MaxPooling2D(pool_size=(3, 3)))
model.add(Dropout(0.25))
model.add(Conv2D(64, (3, 3), padding="same"))
model.add(Activation("relu"))
model.add(BatchNormalization(axis=chanDim))
model.add(Conv2D(64, (3, 3), padding="same"))
model.add(Activation("relu"))
model.add(BatchNormalization(axis=chanDim))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(128, (3, 3), padding="same"))
model.add(Activation("relu"))
model.add(BatchNormalization(axis=chanDim))
model.add(Conv2D(128, (3, 3), padding="same"))
model.add(Activation("relu"))
model.add(BatchNormalization(axis=chanDim))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(1024))
model.add(Activation("relu"))
model.add(BatchNormalization())
model.add(Dropout(0.5))
model.add(Dense(15))
model.add(Activation("softmax"))
model.summary()
任何人都可以解释。
Model: "sequential_2"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d_5 (Conv2D) (None, 256, 256, 32) 896
_________________________________________________________________
activation_6 (Activation) (None, 256, 256, 32) 0
_________________________________________________________________
batch_normalization_6 (Batch (None, 256, 256, 32) 128
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 85, 85, 32) 0
_________________________________________________________________
dropout_4 (Dropout) (None, 85, 85, 32) 0
_________________________________________________________________
conv2d_6 (Conv2D) (None, 85, 85, 64) 18496
_________________________________________________________________
activation_7 (Activation) (None, 85, 85, 64) 0
_________________________________________________________________
batch_normalization_7 (Batch (None, 85, 85, 64) 256
_________________________________________________________________
conv2d_7 (Conv2D) (None, 85, 85, 64) 36928
_________________________________________________________________
activation_8 (Activation) (None, 85, 85, 64) 0
_________________________________________________________________
batch_normalization_8 (Batch (None, 85, 85, 64) 256
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 42, 42, 64) 0
_________________________________________________________________
dropout_5 (Dropout) (None, 42, 42, 64) 0
_________________________________________________________________
conv2d_8 (Conv2D) (None, 42, 42, 128) 73856
_________________________________________________________________
activation_9 (Activation) (None, 42, 42, 128) 0
_________________________________________________________________
batch_normalization_9 (Batch (None, 42, 42, 128) 512
_________________________________________________________________
conv2d_9 (Conv2D) (None, 42, 42, 128) 147584
_________________________________________________________________
activation_10 (Activation) (None, 42, 42, 128) 0
_________________________________________________________________
batch_normalization_10 (Batc (None, 42, 42, 128) 512
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 21, 21, 128) 0
_________________________________________________________________
dropout_6 (Dropout) (None, 21, 21, 128) 0
_________________________________________________________________
flatten_1 (Flatten) (None, 56448) 0
_________________________________________________________________
dense_1 (Dense) (None, 1024) 57803776
_________________________________________________________________
activation_11 (Activation) (None, 1024) 0
_________________________________________________________________
batch_normalization_11 (Batc (None, 1024) 4096
_________________________________________________________________
dropout_7 (Dropout) (None, 1024) 0
_________________________________________________________________
dense_2 (Dense) (None, 15) 15375
_________________________________________________________________
activation_12 (Activation) (None, 15) 0
=================================================================
Total params: 58,102,671
Trainable params: 58,099,791
Non-trainable params: 2,880
_________________________________________________________________
我无法计算卷积层和密集层的数量。我也试试model.layers
。这个的输出是28。怎么样?
如何以编程方式获取卷积层和密集层的数量?
解决方案
首先,层数之所以为 28 是因为Flatten
, BatchNormalization
, Dropout
,Activation
都MaxPool2D
计入model.layers
.
话虽如此,您可以使用以下方法获取层数isinstance
:
num_conv = 0
num_dense = 0
for layer in model.layers:
if isinstance(layer, Conv2D):
num_conv += 1
elif isinstance(layer, Dense):
num_dense += 1
推荐阅读
- opentbs - 我可以使用 TinyButStrong 字段来控制 Word 表中的段落格式吗
- javascript - Angular 7 - *ngFor 没有正确迭代并且没有考虑解构对象
- sql-server - 免费试用后 Azure SQL Server 登录失败
- r - R ggplot boxplot 通过组合因素进行分组
- python - Python dict - 如何将变量设置为值?
- python - 使用 Selenium(Python) 将 Youtube 质量降低到 144p
- javascript - 如何使用 Fuse.js 过滤 vuetify v-autocomplete 组件中的数据?
- php - 使用服务帐户和 PHP 上传 Google 驱动器文件
- jquery-ui - 如何将对象拖放到 three.js Mesh 中?
- postgresql - 是否可以为 Select 语句创建存储过程?