FID使⽤（FrechetInceptionDistancescore）
300m无线路由器何为 FID？
Frechet Inception 距离（FID）是评估⽣成图像质量的度量标准，专门⽤于评估⽣成对抗⽹络的性能。
该分数作为对现有的 Inception 分数（IS）的改进⽽被提出。
介绍：
Frechet Inception 距离得分（Frechet Inception Distance score，FID）是计算真实图像和⽣成图像的特征向量之间距离的⼀种度量。
FID 从原始图像的计算机视觉特征的统计⽅⾯的相似度来衡量两组图像的相似度，这种视觉特征是使⽤ Inception v3 图像分类模型计算的得到的。分数越低代表两组图像越相似，或者说⼆者的统计量越相似，FID 在最佳情况下的得分为 0.0，表⽰两组图像相同。
FID 分数被⽤于评估由⽣成性对抗⽹络⽣成的图像的质量，较低的分数与较⾼质量的图像有很⾼的相关性。
慈母情深教学反思在本教程中，你将了解如何通过 FID 评估⽣成的图像。
风口浪尖⾸先安装：pip install pytorch-fid == 0.1.1
什么叫涨停Usage
To compute the FID score between two datats, where images of each datat are contained in an individual folder:
python -m pytorch_fid path/to/datat1 path/to/datat2
To run the evaluation on GPU, u the flag --device cuda:N, where N is the index of the GPU to u.（现在变成--gpu 0）
例如命令：python -m pytorch_fid groundtruth input --gpu 0
如果要使⽤CPU，直接把--gpu 0删除就可以了，直接 python -m pytorch_fid groundtruth input
Using different layers for feature maps
In difference to the official implementation, you can choo to u a different feature layer of the Inception network instead of the default pool3 layer. As the lower layer features still have spatial exte
nt, the features are first global average pooled to a vector before estimating mean and covariance.
This might be uful if the datats you want to compare have less than the otherwi required 2048 images. Note that this changes the magnitude of the FID score and you can not compare them against scores calculated on another dimensionality. The resulting scores might also no longer correlate with visual quality.
You can lect the dimensionality of features to u with the flag --dims N, where N is the dimensionality of features. The choices are:
64: first max pooling features
192: cond max pooling features
湖南四大名酒768: pre-aux classifier features
2048: final average pooling features (this is the default)
使⽤：
例如在该项⽬下使⽤该命令：python -m pytorch_fid datats/colorization/sidd/val/groundtruth datats/colorization/sidd/val/input --gpu 0
河道清淤工程纷字成语最终得到得分：
密铺图形
