search

cpf0079 / UCDA

Source codes for "Unsupervised Curriculum Domain Adaptation for No-Reference Video Quality Assessment"
17 stars 3 forks source link

could you show the train dataset format, such as the source.txt, source.xls, target.txt, target.xls？ #2

Closed zsrluminous closed 2 years ago

cpf0079 commented 2 years ago

Of course. What are recorded in source.txt/target.txt are their numbers in the datasets, such as 1,2,3,4,... And their corresponding probability distributions of qulity are recored in source.xls/target.xls, such as line 1(no. 1): 0.1,0.3,0.2,0.3,0.1. These five numbers stand for the probabilities of score 1, 2, 3, 4 and 5.

zsrluminous commented 2 years ago

ok, i have another question, i download the LBVD dataset, and the given LBVD_SUBJECTIVEDATA.CSV provide the mos value and varience like this : 3.96 | 0.195959 3.28 | 0.53066

how can i get the labels of quality rating distributions saied in your paper?

cpf0079 commented 2 years ago

I have uploaded the raw_score.xlsx in the repository, you can get the distribution information from it.