search

jinyeying / nighttime_dehaze

[ACMMM2023] "Enhancing Visibility in Nighttime Haze Images Using Guided APSF and Gradient Adaptive Convolution", https://arxiv.org/abs/2308.01738
147 stars 10 forks source link

about the training way #3

Closed diadestiny closed 11 months ago

diadestiny commented 11 months ago

Hello, may I know if your network was trained using an unpaired or paired approach? In your paper, you mentioned that your network was trained based on a CycleGAN-based network (an unpaired training method). However, in the experimental section, you stated that a paired GTA5 dataset was used for training.

jinyeying commented 11 months ago

Hello, may I know if your network was trained using an unpaired or paired approach? In your paper, you mentioned that your network was trained based on a CycleGAN-based network (an unpaired training method). However, in the experimental section, you stated that a paired GTA5 dataset was used for training.

Hi, it is semi-supervised. For the real images do not have ground truth, it is unsupervised. As for the paired GTA5 dataset, it is supervised. The whole method is semi-supervised.