search

chaimi2013 / 3R

Nighttime Dehazing with a Synthetic Benchmark (ACM MM 2020)
54 stars 3 forks source link

ND-Net code #11

Closed chenhuiphd closed 2 years ago

chenhuiphd commented 2 years ago

Hello, did you use 8073 pairs of NHR images to do ND-Net training, and then use 897 haze images to do network testing to calculate PSNR and SSIM parameter indicators? Currently, there is color bias problem in my dehazing result, could you please public your ND-Net code, I would like to learn the parameter configuration below, thank you.

chaimi2013 commented 2 years ago

Hello, did you use 8073 pairs of NHR images to do ND-Net training, and then use 897 haze images to do network testing to calculate PSNR and SSIM parameter indicators? Currently, there is color bias problem in my dehazing result, could you please public your ND-Net code, I would like to learn the parameter configuration below, thank you.

ND-Net is based on a simple encoder-decoder structure, which has been described in the paper. Please refer to the structure details and implement it. By the way, I have provided the download links to NHR and NHC. You can try to train ND-Net on them.