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
890 stars 198 forks source link

What is dataset_mode ?? #101

Closed edwardcho closed 2 years ago

edwardcho commented 2 years ago

Hello Sir.

I saw dataset_mode in your code (data/custom_dataset_data_loader.py). What is dataset_mode??

According to dataset_mode, what should I check for training using my-datasets ? (Actually, I only have low-light images)

Thanks. Edward Cho.

yifanjiang19 commented 2 years ago

You can not train it if you only have low-light images

edwardcho commented 2 years ago

Hello yueruchen.

I think that ElightenGAN is unsupervised learning type for image enhancement. So I think that this network needs only low-light images for training.

If not, what should I prepare training? (If so, this network is not type of unsupervised learning ?)

Thanks, Edward Cho.

yifanjiang19 commented 2 years ago

This is unsupervised learning, but no PAIRED training data. Which means you don't need to provide a normal light ground truth for low-light input images. So you should prepare both the low-light and normal light image examples, but they don't need to be the same scene.