Closed jessiZO closed 3 years ago
Hey @jessiZO ,
Sorry for the delay but we have been pretty busy in this period. For information regarding the experimental setup you can refer to the paper we have presented at ICPR2020. Answering your questions: we used 32 frames per video both in training and in testing, so to classify a whole video we consider the scores extracted from 32 frames selected equally spaced in time. Bests,
Edoardo
So you took average of the scores obtained from those 32 frames?
On Thu, 9 Sep 2021, 8:07 pm Edoardo Daniele Cannas, < @.***> wrote:
Hey @jessiZO https://github.com/jessiZO ,
for information regarding the experimental setup you can refer to the paper https://ieeexplore.ieee.org/document/9412711 we have presented at ICPR2020. Answering your questions: we used 32 frames per video both in training and in testing, so to classify a whole video we consider the scores extracted from 32 frames selected equally spaced in time. Bests,
Edoardo
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/polimi-ispl/icpr2020dfdc/issues/63#issuecomment-916132028, or unsubscribe https://github.com/notifications/unsubscribe-auth/AVLFMOTJV4JQSDRH3QDOTY3UBC5SJANCNFSM5CZXEZMQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
Yes, we took the average of the pre-sigmoid scores and then we computed the sigmoid on the mean score of the whole video :)
The split procedure that you followed for celeb df dataset, will it handle the imbalance in the dataset?
On Thu, 9 Sep 2021, 8:27 pm Edoardo Daniele Cannas, < @.***> wrote:
Yes, we took the average of the pre-sigmoid scores and then we computed the sigmoid on the mean score of the whole video :)
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/polimi-ispl/icpr2020dfdc/issues/63#issuecomment-916149118, or unsubscribe https://github.com/notifications/unsubscribe-auth/AVLFMOQNXVZVFQ42SB2BDFDUBC73NANCNFSM5CZXEZMQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
Hey. I have a question. Have you used all the frames extracted from each video to identify a real/fake video? I tried looking for it but since I'm a starter I am facing difficulties in understanding the code. One more thing, you have used the scores obtained from each frame to classify the whole video right?