The research is well designed and written. It contributes to the development of a strong and user-friendly AI tool that can provide quick and effective support in flood-affected areas where urgent assistance is needed, without requiring harmonized or standardized procedures for image collection from different sources. As a limitation of the research, I believe it would be valuable to suggest including the geolocation of the final flood map to facilitate relief efforts. Furthermore, the reasons behind the superiority of SAM-Points should be discussed. Compared to other methods, this approach appears to be more effective in distinguishing bare soil from flooded areas.

Upon re-reading the manuscript, I noticed that in lines 200–203 you mention the use of various data augmentation techniques. Could you please clarify the probability settings assigned to each augmentation method?



In lines 201–203, it is not clear whether the augmentation was applied exclusively to the training dataset. Providing this clarification would enhance the transparency of the methodology.



Still in lines 201–203, it would be highly valuable to explicitly include details regarding the number of images before and after data augmentation, as well as their distribution across the training, validation, and test sets. Such information is critical to ensure reproducibility.



In lines 209–219, you mention the use of both Dice Loss and Cross-Entropy Loss. Could you please specify how these two loss functions were combined? For example, were they summed, averaged, or weighted differently?



I appreciate that the code is publicly available on GitHub. However, I could not locate the corresponding datasets in the repository. Based on the README file, it seems that the authors expect users to obtain the data from an external source. While this is acceptable provided that the source remains reliably available, hosting a copy of the datasets within your GitHub repository would be preferable for long-term accessibility.

In their paper, Hadi Shokati et al. Propose methods to improve rapid flood mapping from Aerial Imagery using Fine-Tuned SAM and ResNet-Backboned U-Net. This paper is a valuable contribution to remote sensing and rapid disaster assessment. Although none of the comments and suggestions are critical, I would like to ask authors to incorporate and address these issues and suggestions before the paper's publication.

1. Although the methods in this paper are related to floods, they do not directly discuss flood itself. Therefore, it would be beneficial for the readers and also enhance the paper’s visibility to replace “flood” in keywords with “flood mapping” or ”segmentation of flood”, which are more relevant to the presented study.
2. The introduction and methods sections are well written, addressing the main issues and research question. However, there was a minor absence of the reference in line 210 regarding the choice of DiceCELoss. It is claimed in the paper that: “DiceCELoss is often used to improve segmentation performance by leveraging both the pixel-wise accuracy (via Cross-Entropy) and the structural similarity (via Dice coefficient).” Please include at least one reference to support this claim and the choice of this loss function.
3. Although the terms and names of the methods are well described throughout the paper, their usage in the text and figures is inconsistent. For example, Segment Anything Mode is consistently abbreviated as SAM, but the versions (point prompts or points prompts) are referred to in various inconsistent forms. Please use consistent terminology for methods in the entire manuscript, especially for the main methods. For instance, here are a few examples:
   • SAM (Points prompts) on page 14 and SAM (Point prompts) on page 15 in the figures.
   • Points in Figure 4.
   • point prompts in line 309
   • point prompt in line 100
   • "Bounding boxes" is abbreviated as "Bbox" in line 135, but this is inconsistently used throughout the text and figures, sometimes as "bounding box" and other times as "Bbox."
   • ...
4. Figure 6 lacks a brief description of the subplots labeled a, b, ..., h in the caption.
5. I would like to ask the authors to elaborate on why 290 images with different geographic regions and diverse flood events are sufficient for this study. We recognize that transfer learning enables us to train our models with a limited sample size by leveraging pre-trained data; I would appreciate a discussion on how this sample size captures the variability needed for a robust model. Including this clarification would strengthen the manuscript by addressing potential concerns about the dataset.