Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

about result #16

Open
liujiang137 opened this issue Dec 20, 2018 · 7 comments
Open

about result #16

liujiang137 opened this issue Dec 20, 2018 · 7 comments

Comments

@liujiang137
Copy link

I want to know the result score is more high if the picture is more clear.What is the range of the score interval? (when I run the code,there is many matter about xp. so I replace xp with np,This does not affect the results?)

@wunderwuzzi1975
Copy link
Collaborator

wunderwuzzi1975 commented Jan 7, 2019

Happy New Year, @liujiang137 !

I want to know the result score is more high if the picture is more clear.What is the range of the score interval?

According to the paper, "result scores lie in the range [0, 100]. A higher value indicates higher quality., where a lower score indicates better visual image quality." " Please have a look at the paper for more details.

When I run the code,there is many matter about xp. so I replace xp with np,This does not affect the results?

What kind of errors/warnings are you experiencing?

@violivei
Copy link

violivei commented Jan 21, 2019

"Resulting DMOS quality ratings lie in the range of [0, 100], where a lower score indicates better visual image quality."
No?

tbf, it makes more sense based on the results that I'm having here.

@wunderwuzzi1975
Copy link
Collaborator

wunderwuzzi1975 commented Jan 22, 2019

@violivei
Your results are right. I got confused by the different scores used for the different databases. On page 6, right column, last paragraph it is stated that "to make errors and gradients comparable for different databases, the MOS values of TID2013 and CLIVE and the DMOS values of CSIQ have been linearly mapped to the same range as the DMOS values in LIVE". And the DMOS values in LIVE lie in the range of [0, 100], where a lower score indicates better visual image quality. I have edited my posting above to avoid any further confusion.

@chungyau97
Copy link

chungyau97 commented Apr 1, 2019

@wunderwuzzi1975
1.jpg
1
2.jpg
2

The result I get are:
Patchwise
The 1st result is 1.jpg
The 2nd result is 2.jpg
image

Weighted
The 1st result is 1.jpg
The 2nd result is 2.jpg
image

@littleSpongebob
Copy link

@wunderwuzzi1975
excuse me, when i use the pretrain model(live dataset pretrain model) to test some picture ,the model will return 0.0034 or -0.003(return the number very close to zero,), could you tell me what is the problem?
thanks

@ladevan
Copy link

ladevan commented Sep 18, 2020

also got negative results similar to @littleSpongebob

@Ffmydy
Copy link

Ffmydy commented Nov 16, 2020

@chungyau97 ,my result is the same with you,but I can't understand the quality of the 1st should be better,why the score is highter?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

7 participants