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
I have a question about some ambiguity during training model.
According to the paper, when resnet150 is trained, a 300x300 crop is performed centered on the detected nose coordinates. Am I correct in my understanding?
If you do this, the aspect of the input video will be different depending on the distance between the camera and the person.
Have you tried resizing the detected face area to 300x300?
Thank you.
The text was updated successfully, but these errors were encountered:
Hello!
The image position of the nose was used as the center of the crop. I confirm that your understanding is correct.
The distance between the camera and the person is fixed in the dataset. Face resizing was not applied.
Cheers!
Hello. Thanks a lot for your wonderful research.
I have a question about some ambiguity during training model.
According to the paper, when resnet150 is trained, a 300x300 crop is performed centered on the detected nose coordinates. Am I correct in my understanding?
If you do this, the aspect of the input video will be different depending on the distance between the camera and the person.
Have you tried resizing the detected face area to 300x300?
Thank you.
The text was updated successfully, but these errors were encountered: