search

VITA-Group / EnlightenGAN

[IEEE TIP] "EnlightenGAN: Deep Light Enhancement without Paired Supervision" by Yifan Jiang, Xinyu Gong, Ding Liu, Yu Cheng, Chen Fang, Xiaohui Shen, Jianchao Yang, Pan Zhou, Zhangyang Wang
Other
893 stars 198 forks source link

cuda out of memory when predicting #11

Closed Theodoric008 closed 4 years ago

Theodoric008 commented 5 years ago

its ok when training, but cuda out of memory when predicting, how to adjust paramters?

Theodoric008 commented 5 years ago

by the way, i got some result, and it is absolutely IMPRESSIVE

yifanjiang19 commented 5 years ago

Hi, thanks for following our work. Actually, I used a very wide model which is not necessary. if you want to use it for high-resolution image you could try to train a small-size model.

yifanjiang19 commented 5 years ago

Or resize the input image to a low-resolution one and use guided image filtering (http://kaiminghe.com/publications/eccv10guidedfilter.pdf) to get a high-resolution output.

UtopiaHu commented 5 years ago

I met the same problem when I used your provided model to test on some images of your provided testing datasets. How can I make some adjustments if I have only one 1080ti GPU card?

yifanjiang19 commented 5 years ago

@UtopiaHu Try to resize the input images to half size.