GANimation

GANimation copied to clipboard

For Training the ganimation , we resized all images in emotionet dataset to 128 x 128 and cleaned the image as mentioned in paper. After that we passed the image...

sauradip

Output Problem

6

Hello,first I use 1080ti to train 200000 images,everything is ok.The output looks not bad, but also not very good.   So,I retrained 500000 images,all the output looks so bad,but...

Masakaa

How to reproduce the result on Rafd dataset in the paper?

1

Hi @albertpumarola , The visual results on RAFD dataset are quit good. So, did you first train your model on EmotionNet and the test on RAFD dataset? Or did you...

o0t1ng0o

Could you share your data list ?

2

You do an impressive work in emotion transfer. I trained the model with 20k random samples from EmotioNet, but the performance is not as good as yours. I found the...

hao-qiang

EmotionNet

2

Can you share me with the dataset EmotionNet used in your paper? Thanks!

kangwang17

Hi Albert, Very nice job!! I got a question about the loss for AU. When computing the loss for Discriminator's AU regression head, you divided it by batch_size, as shown...

kgu3

First of all, this is a great work, but I have problem about train_G_every_n_iterations. Why do you update G for every 5 steps? Does it mean G is stronger than...

zachluo

lambda parameters for training

1

Hi， I only have 20，000 training images， after training 30 epochs using default parameters， I got the results as following， （it seems generator has no effect）。  I tried...

Jpchen82

About loss

1

Hi, it seems that you ignore the fake image condition loss: E[||D(G(I|y)) - y||^2], is this better for training? ``` # D(fake_I) d_fake_desired_img_prob, **_** = self._D.forward(fake_imgs_masked.detach()) self._loss_d_fake = self._compute_loss_D(d_fake_desired_img_prob, False)...

joyyang1997

I am running this code on the Mac-book-air. During the execution of the following command python test.py --gpu_ids -1 --input_path sample_dataset/imgs/N_000000.jpg I am getting the following error message. I am...

mayanktiwariiiitdmj