In your NIPS paper, you mentioned you use mirror augmentation to expand disney dataset from 300 images to 32 thousand images.
Could you share how you do that?
Because flip image left and right only increased dataset around 2 times.
Moreover, could you share which iteration weight you use?
Thank you for your great work.
In your NIPS paper, you mentioned you use mirror augmentation to expand disney dataset from 300 images to 32 thousand images. Could you share how you do that? Because flip image left and right only increased dataset around 2 times.
Moreover, could you share which iteration weight you use?
looking forward to your reply.