search

imatge-upc / salgan

SalGAN: Visual Saliency Prediction with Generative Adversarial Networks
https://imatge-upc.github.io/salgan
MIT License
368 stars 106 forks source link

Pre-trained Network Parameters #20

Closed faizwhb closed 6 years ago

faizwhb commented 7 years ago

I am doing some experiments on the SALGAN. I noticed that the results reported in the arXiv say that the training is done using 256x192 size images and saliency maps but the code has default values of 320x240. My question is that in the paper you performed training on 256x192 images, which would change the aspect ratio of the SALICON dataset. Isn't that a problem ??

junting commented 6 years ago

Hi @fwahhab89,

The image is smaller in size but the aspect ratio is not changed.

Best, Junting