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
Would it be possible to share the dataset used for dlib_face_recognition_resnet_model_v1.dat . I understand that some images came from VGG and others from FaceScrub, but then others were scrapped from the Internet.
The reason I am asking this is we need to evaluate and prove there is no gender, race or age bias in the model, and having the dataset used for training would help a lot in this.
Thank you,
Joao
The text was updated successfully, but these errors were encountered:
The data isn't available. But you don't need to check for this, the model
is definitely biased towards generally good looking american
celebrities (e.g. white middle aged adults)
On Tue, Nov 3, 2020 at 5:47 AM jpsalada ***@***.***> wrote:
Hi all,
Would it be possible to share the dataset used for
*dlib_face_recognition_resnet_model_v1.dat* . I understand that some
images came from VGG and others from FaceScrub, but then others were
scrapped from the Internet.
The reason I am asking this is we need to evaluate and prove there is no
gender, race or age bias in the model, and having the dataset used for
training would help a lot in this.
Thank you,
Joao
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#42>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABPYFR54XJYHX6VWIJKUROTSN7NSNANCNFSM4TISUIRA>
.
Hi all,
Would it be possible to share the dataset used for dlib_face_recognition_resnet_model_v1.dat . I understand that some images came from VGG and others from FaceScrub, but then others were scrapped from the Internet.
The reason I am asking this is we need to evaluate and prove there is no gender, race or age bias in the model, and having the dataset used for training would help a lot in this.
Thank you,
Joao
The text was updated successfully, but these errors were encountered: