Great work! I would like to ask a simple question about ridcp_dehazing

[RQ-Wu]

Our new data can boost the performance more than the new network. However, the VQ can also improve the visual quality. Besides, we think the most interesting point for VQ is that the codebook brings the ability to adjust the enhancement degree, which can not be achieved by preview learn-based approaches.
Additionally, We didn't evaluate the quantitative results of other methods trained on our data. You can evaluate them by yourself. We used IQA-Pytorch for evaluation.

[IDKiro]

Thansk, I may try it.