728x90
반응형
1. LeNet-5 코드
1-1 LeNet-5 코드 function api
from tensorflow.keras.layers import Conv2D,Dense,Flatten,Input,AveragePooling2D
from tensorflow.keras import Model
image_input = Input(shape=(32, 32, 1))
x = Conv2D(6, kernel_size= (5,5), strides=1, activation = 'relu')(image_input)
x = AveragePooling2D(2)(x)
x = Conv2D(16, kernel_size= (5,5),strides=1, activation = 'relu')(x)
x = AveragePooling2D(2)(x)
x = Flatten()(x)
x= Dense(units = 120, activation = 'relu')(x)
x= Dense(units = 84, activation = 'relu')(x)
output = Dense(10, activation='softmax')(x)
model = Model(image_input, output)
model.summary()
1-2 LeNet-5
import keras
from keras.models import Sequential
from keras.layers import Conv2D
from keras.layers import MaxPooling2D
from keras.layers import Flatten
from keras.layers import Dense
model = Sequential()
#Layer 1
#Conv Layer 1
model.add(Conv2D(filters = 6, kernel_size = 5, strides = 1, activation = 'relu', input_shape = (32,32,1)))
#Pooling layer 1
model.add(MaxPooling2D(pool_size = 2, strides = 2))
#Layer 2
#Conv Layer 2
model.add(Conv2D(filters = 16, kernel_size = 5,strides = 1,activation = 'relu'))
#Pooling Layer 2
model.add(MaxPooling2D(pool_size = 2, strides = 2))
#Flatten
model.add(Flatten())
#Layer 3
#Fully connected layer 1
model.add(Dense(units = 120, activation = 'relu'))
#Layer 4
#Fully connected layer 2
model.add(Dense(units = 84, activation = 'relu'))
#Layer 5
#Output Layer
model.add(Dense(units = 10, activation = 'softmax'))
model.summary()
AlexNet keras 코드
# 5 convolutional laysers, 3 maxpooling , 3 fully connected
import keras
from keras.models import Sequential
from keras.layers import Convolution2D
from keras.layers import MaxPooling2D,Dropout
from keras.layers import Flatten
from keras.layers import Dense
model = Sequential()
model.add(Convolution2D(96, (11, 11), strides=(4, 4), input_shape = (224,224,3),padding='SAME', activation='relu'))
model.add(MaxPooling2D((3,3) , strides = 2))
model.add(Convolution2D(256, (5, 5), padding = 'same', activation='relu'))
model.add(MaxPooling2D((3,3) , strides = 2))
model.add(Convolution2D(384, (3, 3), padding = 'same', activation='relu'))
model.add(Convolution2D(384, (3, 3), padding = 'same', activation='relu'))
model.add(Convolution2D(256, (3, 3), padding = 'same', activation='relu'))
model.add(MaxPooling2D((3,3) , strides = 1))
model.add(Flatten())
model.add(Dense(4096, activation = 'relu'))
model.add(Dropout(0.5))
model.add(Dense(4096, activation = 'relu'))
model.add(Dropout(0.5))
model.add(Dense(1000, activation = 'softmax'))
model.summary()
VGG코드
VGG16 tensorlfow 코드
import tensorflow as tf
model = tf.keras.applications.VGG16(
include_top=True, weights='imagenet', input_tensor=None,
input_shape=None, pooling=None, classes=1000,
classifier_activation='softmax'
)
model.summary()block 1 2 1
block 2 2 1
block 3 2 1
block 4 2 1
block 5 3 1
flatten
dense 3
from tensorflow.keras.layers import Conv2D,Dense,Flatten,Input,MaxPooling2D
from tensorflow.keras import Model
image_input = Input(shape=(224, 224, 3))
x = Conv2D(64, kernel_size= (3,3), padding='same',name='block1_conv1')(image_input)
x = Conv2D(64, kernel_size= (3,3), padding='same',name='block1_conv2')(x)
x = MaxPooling2D(2 ,name ='block1_pool')(x)
x = Conv2D(128, kernel_size= (3,3), padding='same',name='block2_conv1')(x)
x = Conv2D(128, kernel_size= (3,3), padding='same',name='block2_conv2')(x)
x = MaxPooling2D(2 ,name ='block2_pool')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv1')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv2')(x)
x = MaxPooling2D(2 ,name ='block3_pool')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv1')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv2')(x)
x = MaxPooling2D(2 ,name ='block4_pool')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv1')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv2')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv3')(x)
x = MaxPooling2D(2 ,name ='block5_pool')(x)
x = Flatten()(x)
x= Dense(units = 4096, activation = 'relu',name='fc1')(x)
x= Dense(units = 4096, activation = 'relu',name='fc2')(x)
output = Dense(1000, activation='softmax' ,name='predictions')(x)
model = Model(image_input, output)
model.summary()www.tensorflow.org/api_docs/python/tf/keras/applications/VGG16
tf.keras.applications.VGG16 | TensorFlow Core v2.4.1
Instantiates the VGG16 model.
www.tensorflow.org
VGG19 tensorlfow 코드
import tensorflow as tf
model = tf.keras.applications.VGG19(
include_top=True, weights='imagenet', input_tensor=None,
input_shape=None, pooling=None, classes=1000,
classifier_activation='softmax'
)
model.summary()from tensorflow.keras.layers import Conv2D,Dense,Flatten,Input,MaxPooling2D
from tensorflow.keras import Model
image_input = Input(shape=(224, 224, 3))
x = Conv2D(64, kernel_size= (3,3), padding='same',name='block1_conv1')(image_input)
x = Conv2D(64, kernel_size= (3,3), padding='same',name='block1_conv2')(x)
x = MaxPooling2D(2 ,name ='block1_pool')(x)
x = Conv2D(128, kernel_size= (3,3), padding='same',name='block2_conv1')(x)
x = Conv2D(128, kernel_size= (3,3), padding='same',name='block2_conv2')(x)
x = MaxPooling2D(2 ,name ='block2_pool')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv1')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv2')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv3')(x)
x = Conv2D(256, kernel_size= (3,3), padding='same',name='block3_conv4')(x)
x = MaxPooling2D(2 ,name ='block3_pool')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv1')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv2')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv3')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block4_conv4')(x)
x = MaxPooling2D(2 ,name ='block4_pool')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv1')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv2')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv3')(x)
x = Conv2D(512, kernel_size= (3,3), padding='same',name='block5_conv4')(x)
x = MaxPooling2D(2 ,name ='block5_pool')(x)
x = Flatten()(x)
x= Dense(units = 4096, activation = 'relu',name='fc1')(x)
x= Dense(units = 4096, activation = 'relu',name='fc2')(x)
output = Dense(1000, activation='softmax' ,name='predictions')(x)
model = Model(image_input, output)
model.summary()ResNet코드
ResNet50 tensorflow code
model = tf.keras.applications.ResNet50(
include_top=True, weights='imagenet', input_tensor=None,
input_shape=None, pooling=None, classes=1000
)
model.summary()
ResNet101 tensorflow code
model = tf.keras.applications.ResNet101(
include_top=True, weights='imagenet', input_tensor=None,
input_shape=None, pooling=None, classes=1000
)
model.summary()ResNet152 tensorflow code
model = tf.keras.applications.ResNet152(
include_top=True, weights='imagenet', input_tensor=None,
input_shape=None, pooling=None, classes=1000
)
model.summary()
Keras documentation: Keras Applications
Keras Applications Keras Applications are deep learning models that are made available alongside pre-trained weights. These models can be used for prediction, feature extraction, and fine-tuning. Weights are downloaded automatically when instantiating a mo
keras.io
resnet 50 tensorflow
def conv_block(input_tensor,
kernel_size,
filters,
stage,
block,
strides=(2, 2)):
bn_axis = 3
filters1, filters2, filters3 = filters
conv_name_base = 'conv'+str(stage)+'_block' + block
bn_name_base = 'conv' + str(stage) +'_block'+ block
relu_name_base = 'conv' + str(stage) +'_block'+ block
x = layers.Conv2D(filters1, (1, 1), strides=strides,
kernel_initializer='he_normal',
name=conv_name_base + '_1_conv')(input_tensor)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_1_bn')(x)
x = layers.Activation('relu',name = relu_name_base+'_1_relu')(x)
x = layers.Conv2D(filters2, kernel_size, padding='same',
kernel_initializer='he_normal',
name=conv_name_base + '_2_conv')(x)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_2_bn')(x)
x = layers.Activation('relu',name = relu_name_base+'_2_relu')(x)
x = layers.Conv2D(filters3, (1, 1),
kernel_initializer='he_normal',
name=conv_name_base + '_0_conv')(x)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_0_bn')(x)
shortcut = layers.Conv2D(filters3, (1, 1), strides=strides,
kernel_initializer='he_normal',
name=conv_name_base + '_3_conv')(input_tensor)
shortcut = layers.BatchNormalization(
axis=bn_axis, name=bn_name_base + '_3_bn')(shortcut)
x = layers.add([x, shortcut] ,name = 'conv'+str(stage)+'_block'+block+'_add')
x = layers.Activation('relu',name = 'conv'+str(stage)+'_block'+block+'_out')(x)
return x
def identity_block(input_tensor, kernel_size, filters, stage, block):
"""The identity block is the block that has no conv layer at shortcut.
# Arguments
input_tensor: input tensor
kernel_size: default 3, the kernel size of
middle conv layer at main path
filters: list of integers, the filters of 3 conv layer at main path
stage: integer, current stage label, used for generating layer names
block: 'a','b'..., current block label, used for generating layer names
# Returns
Output tensor for the block.
"""
filters1, filters2, filters3 = filters
bn_axis = 3
conv_name_base = 'conv' + str(stage) +'_block' + '_'+block
bn_name_base = 'conv' + str(stage) +'_block' + block
relu_name_base = 'conv' + str(stage) +'_block'+ block
x = layers.Conv2D(filters1, (1, 1),
kernel_initializer='he_normal',
name=conv_name_base + '_1_conv')(input_tensor)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_1_bn')(x)
x = layers.Activation('relu',name = relu_name_base+'_1_relu')(x)
x = layers.Conv2D(filters2, kernel_size,
padding='same',
kernel_initializer='he_normal',
name=conv_name_base + '_2_conv')(x)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_2_bn')(x)
x = layers.Activation('relu',name = relu_name_base+'_2_relu')(x)
x = layers.Conv2D(filters3, (1, 1),
kernel_initializer='he_normal',
name=conv_name_base + '_3_conv')(x)
x = layers.BatchNormalization(axis=bn_axis, name=bn_name_base + '_3_bn')(x)
x = layers.add([x, input_tensor],name = 'conv'+str(stage)+'_block'+block+'_add')
x = layers.Activation('relu',name = 'conv'+str(stage)+'_block'+block+'_out')(x)
return x
from tensorflow.keras.layers import Conv2D,Dense,Flatten,Input,MaxPooling2D,ZeroPadding2D,BatchNormalization,Activation
from tensorflow.keras import Model
from tensorflow.keras import layers
image_input = Input(shape=(224, 224, 3))
x = ZeroPadding2D(3,name='conv1_pad')(image_input)
x = Conv2D(64, kernel_size= (7,7), strides=(2, 2),padding='valid',name='conv1_conv')(x)
x = BatchNormalization(axis=3, name='bn_conv1')(x)
x = Activation('relu', name='conv1_relu')(x)
x = ZeroPadding2D(padding=(1, 1),name='pool1_pad')(x)
x = MaxPooling2D((3, 3), strides=(2, 2),name='pool1_pool')(x)
x = conv_block(x, 3, [64, 64, 256], stage=2, block='1', strides=(1, 1))
x = identity_block(x, 3, [64, 64, 256], stage=2, block='2')
x = identity_block(x, 3, [64, 64, 256], stage=2, block='3')
x = conv_block(x, 3, [128, 128, 512], stage=3, block='1')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='2')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='3')
x = identity_block(x, 3, [128, 128, 512], stage=3, block='4')
x = conv_block(x, 3, [256, 256, 1024], stage=4, block='1')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='2')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='3')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='4')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='5')
x = identity_block(x, 3, [256, 256, 1024], stage=4, block='6')
x = conv_block(x, 3, [512, 512, 2048], stage=5, block='1')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='2')
x = identity_block(x, 3, [512, 512, 2048], stage=5, block='3')
x = layers.GlobalAveragePooling2D(name='avg_pool')(x)
#https://github.com/keras-team/keras-applications/blob/master/keras_applications/resnet50.py
x = Flatten()(x)
x= Dense(units = 4096, activation = 'relu',name='fc1')(x)
x= Dense(units = 4096, activation = 'relu',name='fc2')(x)
output = Dense(1000, activation='softmax' ,name='predictions')(x)
model = Model(image_input, output)
print(model.summary())
반응형
'Deep learning > 소스' 카테고리의 다른 글
| yolov5 coco data set training (0) | 2021.09.17 |
|---|---|
| classification mnist-LeNet-5 (0) | 2021.04.03 |
| yolov5 모델pt (0) | 2021.03.26 |
| python 한국 시간으로 설정 (0) | 2020.11.17 |
| 20201110-yolov5로 데이터 학습 (2) | 2020.11.04 |