nikita.Stracke answered on October 8th 19 at 01:37
You may want to delve into the metadata, rather not say.
destin_Miller answered on October 8th 19 at 01:39
And what if you want to build? What is your "quality" images? :)
Ericka.Kling answered on October 8th 19 at 01:41
And if you check characteristic format loss?
For example for the jpeg noise, blur, Squareness.
To train the neural network and forward.
This is all theoretical considerations, I will be glad to comments.
ramon_Block answered on October 8th 19 at 01:43
Generally, "quality" can be linked to a purely mathematical things and looking at the spectrum (or Fourier decomposition by wavelets), and then to check on how "good" the picture (for example, if there is little spatial frequencies, this simple and clear geometry)
Delbert_Bauch answered on October 8th 19 at 01:45
In the exif of the file sometimes there's a Quality field, but the values are there such as Fine, Average — not a lot...