You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
hi @ryanjay0 , thanks for so awesome project. After I compiled(using the same version caffe ,models as this project) and tested, it seems that the model is low accuracy.
eg. most porn image are mis-classified as 'other' class(ie. non-sexual) but in fact they are 'titfuck' or 'blowjob_handjob' [1]
what are the situations behind your training data? what size, full nude, strict definitions of categories .... etc
Pardon me if the image below is looked uncomfortable (but I think it's a must to illustrate the problem mentioned above)
[1]
The text was updated successfully, but these errors were encountered:
hi @ryanjay0 , thanks for so awesome project. After I compiled(using the same version caffe ,models as this project) and tested, it seems that the model is low accuracy.
eg. most porn image are mis-classified as 'other' class(ie. non-sexual) but in fact they are 'titfuck' or 'blowjob_handjob' [1]
what are the situations behind your training data? what size, full nude, strict definitions of categories .... etc
Pardon me if the image below is looked uncomfortable (but I think it's a must to illustrate the problem mentioned above)
[1]
The text was updated successfully, but these errors were encountered: