keras对keras进行的2D CNN回归计算的损失值是否明智? [英] Is the loss value computed by keras for 2D CNN regression by keras point wise?

问题描述

我正在使用keras在2D图像上进行CNN回归，均方误差作为损失函数.损耗值的范围为100.要知道每个像素的平均误差，我应该将其除以像素总数吗?还是显示的损耗值是针对像素的?

I'm using keras for CNN on 2D images for regression with mean squared error as the loss function. The loss values are of the range 100. To know average error at each pixel, should I divide it by total number of pixels? Or the loss values displayed are for pixels?

推荐答案

在Keras中定义mse的方式使它可以计算平均像素误差.因此，您可以简单地将损失值作为平均像素误差.

The way how mse is defined in Keras make it compute an average pixel error. So you could simply take a loss value as an average pixel error.

这篇关于keras对keras进行的2D CNN回归计算的损失值是否明智?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！