Nighttime Image Dehazing Based on Multi-Scale Gated Fusion Network
Round 1
Reviewer 1 Report
· The multi-scale gated method provides very interesting results, with clear details and fairly natural rendering.
· The difference between the daytime and nighttime methods is the use of normalization (NM) as a derived input instead of only gamma correction.
· Watch out for repeated words at the end of the legend of the Figure 4.
· Is it possible to put Figure 4 and Figure 5 closer to their mention in Section 5?
· Zhang, Jing, et al. "Nighttime dehazing with a synthetic benchmark." Proceedings of the 28th ACM international conference on multimedia. 2020. à This work provides synthetic indoor images with colored fog. This makes it possible to test the relevance of the White Balance technique under extreme conditions.
· In Table 3, quantitative evaluation of underwater images is provided. Can you add some qualitative results of this kind of images ?
· Figure 7 : Can you add nighttime images to show the relevance of the gated fusion method ?
· Similar to Figure 7 in [12], can you show the effectiveness of the multi-scale method in this manuscript but on nighttime images ?
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents a deep learning-based dehazing framework, which adopts a fusion-based strategy using three inputs from an original input by applying White Balance, Contrast Enhancing, and Gamma Correction.
Overall, the paper is well written and the proposed approach is evaluated in extensive experimental results.
1. It seems that three versions of the model are illustrated in Figure 2. What is the difference between these three (versions or models)?
2. What is the intuition/consideration for designing different models to address the "daytime" dehazing model and the "nighttime" dehazing model?
3. Line 174: It might be better to illustrate the encoder-decoder model used in Figure 2.
4.Figures 5 and 6: It is a bit difficult to see the visual difference among these results. A close-up view might be needed to zoom in on certain regions.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Thank you for taking my comments into considerationAccept in present form
