Question

我在后端Tensorflow上使用Keras。我从（https://github.com/GeorgeSeif/VDSR-Keras）下载了非常深的超分辨率cnn的代码，并尝试训练自己的模型。该模型使用均方误差（mse）进行编译，但是原始模型具有残差。如何在喀拉拉邦实现剩余损失。剩余损失可以计算为

loss = (label - input_image - predected_image)^2

Answer 1

您可以创建自定义损失函数并进行分配。我不确定残留损失的工作原理，但是如果您能够计算出来，您可以像这样实现。

<?php
$gendate = new DateTime();
$gendate->setISODate(2019,1,3); //year , week num , day
echo $gendate->format('Y-m-d'); //"prints"  2019-01-02

Answer 2

VDSR模型如下：

input_img = Input(shape=IMG_SIZE)

model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(input_img)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)

model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)

model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)

model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(64, (3, 3), padding='same', kernel_initializer='he_normal')(model)
model = Activation('relu')(model)
model = Conv2D(1, (3, 3), padding='same', kernel_initializer='he_normal')(model)
res_img = model

output_img = add([res_img, input_img])

model = Model(input_img, output_img)

模型的输入图像是一块图像块（一个块的分辨率为41 x 41像素，每像素8位）。
在您的公式中： loss =（label-input_image-predected_image）^ 2 =（label-（input_image + predected_image））^ 2）其中input_image + predected_image是newtork输出。网络学习对缩放后的输入图像进行校正（线output_img = add（[res_img，input_img]）），以克服双三次插值的影响，这不是图像超分辨率的最佳方法。

Keras中VDSR的残余损耗的计算

2 个答案: