Open happsky opened 6 years ago
I have a similar question (and issue). By hand, I picked only the identities used in the figure in the main page of this repository ("Results 2. Facial Expression Synthesis on RaFD"). These ID's are Rafd09058, Rafd09054, Rafd09036, Rafd09033,Rafd09031,Rafd09023,Rafd04568* (three images per ID in each facial expression). So, my test split is even less than 10% of the whole dataset (4824 images, only frontal and half frontal ones, 45 and 90 deg). I applied SFD face detector and get square patches by keeping images' original aspect ratios (as in figures).
I trained with standard parameters given here, however generated images at the end of 200K iterations are far from the figure here and also papers supplementary material. Is there any difference between the ones that you generate these images and repositories current parameters (i.e. learning/solver parameters or networks)?
I think you should train 200 epochs instead of 200K iterations.
Can you share your rafd data?
Could you share you rafd dataset ,please?
They have a website for the database. You should not request to share it from another user. It is not ethically correct and also illegal to distribute (because they have an user agreement). Here is the link: http://www.socsci.ru.nl:8180/RaFD2/RaFD?p=main
Could you plz share the dataset link?
Hi~@omersumer,How many days have you waited for the permission of the RaFD dataset ?I have waited for a week and have not received the approval....
Hi @fanwl123, I registered on the website and they confirmed and gave access to download in a few days.
hello could you share me with the RaFD dataset by mailbox:1647560307@qq.com i am Anxious to do graduation design for undergraduates,thanks very much!!!
Hi @fanwl123, I registered on the website and they confirmed and gave access to download in a few days.
hello could you share me with the RaFD dataset by mailbox:1647560307@qq.com i am Anxious to do graduation design for undergraduates,thanks very much!!!
Could you sharing training/testing splitting on RaFD dataset of your StarGAN paper? Moreover, I download RaFD which has 8,040 images, while in your paper you reported it only has 4,824 images, may you also provide the images you used in your paper?