"""
Created on 2021/3/15 9:58.
@Author: haifei
"""
from tensorflow.keras.layers import Input, Conv2D, BatchNormalization, Activation
from tensorflow.keras.layers import UpSampling2D, ZeroPadding2D, DepthwiseConv2D
from tensorflow.keras import Model
def _conv_block(inputs, filters, alpha, kernel=(3, 3), strides=(1, 1)):
filters = int(filters * alpha)
# conv + bn + relu <-- yolo v3 darknet
x = ZeroPadding2D(padding=(1, 1))(inputs)
x = Conv2D(filters, kernel, padding='valid', strides=strides)(x)
x = BatchNormalization()(x)
x = Activation("relu")(x)
return x
def _depthwise_conv_block(inputs, pointwise_conv_filters, alpha, depth_multiplier=1, strides=(1, 1)):
pointwise_conv_filters = int(pointwise_conv_filters * alpha)
x = ZeroPadding2D((1, 1))(inputs)
x = DepthwiseConv2D((3, 3), padding='valid', depth_multiplier=depth_multiplier, strides=strides)(x)
x = BatchNormalization()(x)
x = Activation("relu")(x)
x = Conv2D(pointwise_conv_filters, (1, 1), padding='same', strides=(1, 1))(x)
x = BatchNormalization()(x)
x = Activation("relu")(x)
return x
def get_mobilenet_encoder(inputs): # mobilenetv1=mobilenets
alpha = 1.0
depth_multiplier = 1
x = _conv_block(inputs, 32, alpha, strides=(2, 2)) # 下采样
x = _depthwise_conv_block(x, 64, alpha, depth_multiplier)
f1 = x
x = _depthwise_conv_block(x, 128, alpha, depth_multiplier, strides=(2, 2)) # 下采样
x = _depthwise_conv_block(x, 128, alpha, depth_multiplier)
f2 = x
x = _depthwise_conv_block(x, 256, alpha, depth_multiplier, strides=(2, 2)) # 下采样
x = _depthwise_conv_block(x, 256, alpha, depth_multiplier)
f3 = x
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier, strides=(2, 2)) # 下采样
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier)
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier)
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier)
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier)
x = _depthwise_conv_block(x, 512, alpha, depth_multiplier)
f4 = x
x = _depthwise_conv_block(x, 1024, alpha, depth_multiplier, strides=(2, 2))
x = _depthwise_conv_block(x, 1024, alpha, depth_multiplier)
f5 = x
return [f1, f2, f3, f4, f5]
def get_segnet_decoder(feature):
#
x = Conv2D(512, (3, 3), strides=1, padding='same', activation='relu')(feature)
x = BatchNormalization()(x)
#
x = UpSampling2D(size=(2, 2))(x)
x = Conv2D(256, (3, 3), strides=1, padding='same', activation='relu')(x)
x = BatchNormalization()(x)
#
x = UpSampling2D(size=(2, 2))(x)
x = Conv2D(128, (3, 3), strides=1, padding='same', activation='relu')(x)
x = BatchNormalization()(x)
#
x = UpSampling2D(size=(2, 2))(x)
x = Conv2D(128, (3, 3), strides=1, padding='same', activation='relu')(x)
x = BatchNormalization()(x)
#
x = UpSampling2D(size=(2, 2))(x)
x = Conv2D(64, (3, 3), strides=1, padding='same', activation='relu')(x)
x = BatchNormalization()(x)
#
return x
def build_model(tif_size, bands, class_num):
from pathlib import Path
import sys
print('===== %s =====' % Path(__file__).name)
print('===== %s =====' % sys._getframe().f_code.co_name)
# 输入
inputs = Input(shape=(tif_size, tif_size, bands))
# 编码器
levels = get_mobilenet_encoder(inputs)
# 解码器
x = get_segnet_decoder(feature=levels[3])
# 输出
x = Conv2D(class_num, (1, 1), strides=1, padding='same', activation='softmax')(x)
mymodel = Model(inputs, x)
return mymodel