Y-net

Y-NET: MULTI-SCALE FEATURE AGGREGATION NETWORK WITH WAVELET STRUCTURE SIMILARITY LOSS FUNCTION FOR SINGLE IMAGE DEHAZING - ICASSP 2020

This repository shows implementation of Y-net and Wavelet Structure Simlarity Loss Function.

We provide the loss function and the Y-net. Please refer our previous repository that contains wholes training codes.

Dependencies

• Python 3.6+
• PyTorch 1.1.0+
• PyWt

Usage

1. Cloning the repository

$ git clone https://github.com/dectrfov/Y-net.git
$ cd Y-net

2. Testing

$ python demo.py -d haze_img -m model/model_last.pkl

3. Use the Wavelet Structure Simlarity Loss Function

from wavelet_ssim_loss import WSloss
loss = WSloss()