the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 May 2023
Eye of Horus: A Vision-based Framework for Real-time Water Level Measurement
Abstract. Heavy rains and tropical storms often result in floods, which are expected to increase in frequency and intensity. Flood prediction models and inundation mapping tools provide decision-makers and emergency responders with crucial information to better prepare for these events. However, the performance of models relies on the accuracy and timeliness of data received from in-situ gaging stations and remote sensing; each of these data sources has its limitations, especially when it comes to real-time monitoring of floods. This study presents a vision-based framework for measuring water levels and detecting floods using Computer Vision and Deep Learning (DL) techniques. The DL models use time-lapse images captured by surveillance cameras during storm events for the semantic segmentation of water extent in images. Three different DL-based approaches, namely PSPNet, TransUNet, and SegFormer, were applied and evaluated for semantic segmentation. The predicted masks are transformed into water level values by intersecting the extracted water edges, with the 2D representation of a point cloud generated by an Apple iPhone 13 Pro LiDAR sensor. The estimated water levels were compared to reference data collected by an ultrasonic sensor. The results showed that SegFormer outperformed other DL-based approaches by achieving 99.55 % and 99.81 % for Intersection over Union (IoU) and accuracy, respectively. Moreover, the highest correlations between reference data and the vision-based approach reached above 0.98 for both the coefficient of determination (R2) and Nash-Sutcliffe Efficiency. This study demonstrates the potential of using surveillance cameras and Artificial Intelligence for hydrologic monitoring and their integration with existing surveillance infrastructure.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(0 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-857', Remy Vandaele, 13 Jun 2023
Thank you for the interesting manuscript. In my review below, I have tried to answer the review criteria mentioned on the journal website.
- Does the paper address relevant scientific questions within the scope of HESS?
Yes. The paper offers a new approach to measure river water levels using cameras and deep learning, and its concrete implementation for a use case. This could be considered as the development of a new source of data to better observe river water levels.
- Does the paper present novel concepts, ideas, tools, or data?
Yes. While the technology they use is not new (but state of the art), the idea to merge iPhone LiDAR data with water segmentation, and propose a concrete implementation with Arduino & Raspberry Pi has not been tested until now, to my knowledge.
- Are substantial conclusions reached?
The scope of this study is limited to a single site and there is no discussion as to how easy the methodology employed in this work could generalize, but the results obtained show good promise.
- Are the scientific methods and assumptions valid and clearly outlined?
Yes. The authors provide a detailed presentation of their methodology and technologies. I would be interested to see more experiments to assert the genericity and ease of use of the method, but it might go beyond the scope of this current work.
The main one I have in mind would be to create a water segmentation network that does not rely on the images of the same camera and compare with the current results. Such datasets exist (see [1] where some of them are cited). In practice, having to annotate hundreds of images at each site to make the method work seems a hard limitation on the scope of this work.
- Are the results sufficient to support the interpretations and conclusions?
Yes, although the conclusions made by the authors remain limited in scope.
- Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?
Yes. The authors also provide a github link with their code. The research is dependent on the proprietary Apple iPhone 13 LiDAR sensor, but it seems reasonable.
- Do the authors give proper credit to related work and clearly indicate their own new/original contribution?
Yes.
- Does the title clearly reflect the contents of the paper?
Yes.
- Does the abstract provide a concise and complete summary?
Yes.
- Is the overall presentation well structured and clear?
Yes.
- Is the language fluent and precise?
Yes.
- Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?
- Equation 1: "camrea" should be "camera"
- The efficiency criteria used in the work (R², NSE, RMSE, PBIAS) should be defined with a formula for more clarity.
- Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?
- L161-170. This part should be clarified. I am not sure how the GCPs are used to register the LiDAR sub-regions that were captured. An explanation of the AruCo marker would help. Maybe simplify here and refer+merge with Section 4.3?
- Figure 1 should be better explained, especially Fig 1b with, I think, the AruCo markers & the white numbers in black background.
- L298-301/Table 2. This should be better motivated or explained. As I understand, I am not sure why non river ground truth pixels should be ignored.
- L287-292. There is no mention of how the KNN K parameter was validated. I also wonder if that parameter played a role in the results (L376-385).
- L316-331. Isn't there a possibility that the Transformer networks are "overfitting" the single camera training set?
- Are the number and quality of references appropriate?
Yes, although I was a bit surprised to not see any mention of [1, 2] as the work is quite related.
[1] Vandaele et al., https://doi.org/10.1007/978-3-030-71278-5_17
[2] Vandaele et al., https://doi.org/10.5194/hess-25-4435-2021
Citation: https://doi.org/10.5194/egusphere-2023-857-RC1 -
AC1: 'Reply on RC1', Erfan Goharian, 09 Aug 2023
The authors extend their sincere gratitude to the esteemed reviewers for dedicating their time and expertise to the assessment of this manuscript. This correspondence serves as a comprehensive account of the meticulous revisions and enhancements undertaken in accordance with their invaluable feedback. In thoughtful response to the insightful comments provided by the reviewers, several refinements have been incorporated into the manuscript. Notably, attached document contains additional elucidations aimed at elucidating and substantiating specific facets of the paper. We wish to reiterate our profound appreciation to the editor and the reviewers for their indispensable contributions, which have undeniably enriched the quality and scope of this work.
-
RC2: 'Comment on egusphere-2023-857', Anonymous Referee #2, 21 Jul 2023
This manuscript presents an original vision-based water level measurement approach applicable to real-time flood conditions. While I'm not able to evaluate many technical aspects relating to computer vision, AI and electronics involved in this technique, I found that the approach is rigorous, the methods are sound and openly available, and the research work is very well presented. I therefore feel that this contribution is significant for the water-related research community. Hereafter are a few comments for discussion and improvement of the paper.
It should be noted somewhere (e.g. in the abstract) that the system cannot work at night.
L142 The geographical coordinates of the site would be useful.
4.3 L266 Is the focal length fixed or variable?
Why don’t you estimate the DLT parameters directly? What is the advantage of estimating the intrinsic parameters beforehand?
Editorial:
Eq. 1: camera (camrea)
Citation: https://doi.org/10.5194/egusphere-2023-857-RC2 -
AC2: 'Reply on RC2', Erfan Goharian, 09 Aug 2023
The authors extend their sincere gratitude to the esteemed reviewers for dedicating their time and expertise to the assessment of this manuscript. This correspondence serves as a comprehensive account of the meticulous revisions and enhancements undertaken in accordance with their invaluable feedback. In thoughtful response to the insightful comments provided by the reviewers, several refinements have been incorporated into the manuscript. Notably, attached document contains additional elucidations aimed at elucidating and substantiating specific facets of the paper. We wish to reiterate our profound appreciation to the editor and the reviewers for their indispensable contributions, which have undeniably enriched the quality and scope of this work.
-
AC2: 'Reply on RC2', Erfan Goharian, 09 Aug 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-857', Remy Vandaele, 13 Jun 2023
Thank you for the interesting manuscript. In my review below, I have tried to answer the review criteria mentioned on the journal website.
- Does the paper address relevant scientific questions within the scope of HESS?
Yes. The paper offers a new approach to measure river water levels using cameras and deep learning, and its concrete implementation for a use case. This could be considered as the development of a new source of data to better observe river water levels.
- Does the paper present novel concepts, ideas, tools, or data?
Yes. While the technology they use is not new (but state of the art), the idea to merge iPhone LiDAR data with water segmentation, and propose a concrete implementation with Arduino & Raspberry Pi has not been tested until now, to my knowledge.
- Are substantial conclusions reached?
The scope of this study is limited to a single site and there is no discussion as to how easy the methodology employed in this work could generalize, but the results obtained show good promise.
- Are the scientific methods and assumptions valid and clearly outlined?
Yes. The authors provide a detailed presentation of their methodology and technologies. I would be interested to see more experiments to assert the genericity and ease of use of the method, but it might go beyond the scope of this current work.
The main one I have in mind would be to create a water segmentation network that does not rely on the images of the same camera and compare with the current results. Such datasets exist (see [1] where some of them are cited). In practice, having to annotate hundreds of images at each site to make the method work seems a hard limitation on the scope of this work.
- Are the results sufficient to support the interpretations and conclusions?
Yes, although the conclusions made by the authors remain limited in scope.
- Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?
Yes. The authors also provide a github link with their code. The research is dependent on the proprietary Apple iPhone 13 LiDAR sensor, but it seems reasonable.
- Do the authors give proper credit to related work and clearly indicate their own new/original contribution?
Yes.
- Does the title clearly reflect the contents of the paper?
Yes.
- Does the abstract provide a concise and complete summary?
Yes.
- Is the overall presentation well structured and clear?
Yes.
- Is the language fluent and precise?
Yes.
- Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?
- Equation 1: "camrea" should be "camera"
- The efficiency criteria used in the work (R², NSE, RMSE, PBIAS) should be defined with a formula for more clarity.
- Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?
- L161-170. This part should be clarified. I am not sure how the GCPs are used to register the LiDAR sub-regions that were captured. An explanation of the AruCo marker would help. Maybe simplify here and refer+merge with Section 4.3?
- Figure 1 should be better explained, especially Fig 1b with, I think, the AruCo markers & the white numbers in black background.
- L298-301/Table 2. This should be better motivated or explained. As I understand, I am not sure why non river ground truth pixels should be ignored.
- L287-292. There is no mention of how the KNN K parameter was validated. I also wonder if that parameter played a role in the results (L376-385).
- L316-331. Isn't there a possibility that the Transformer networks are "overfitting" the single camera training set?
- Are the number and quality of references appropriate?
Yes, although I was a bit surprised to not see any mention of [1, 2] as the work is quite related.
[1] Vandaele et al., https://doi.org/10.1007/978-3-030-71278-5_17
[2] Vandaele et al., https://doi.org/10.5194/hess-25-4435-2021
Citation: https://doi.org/10.5194/egusphere-2023-857-RC1 -
AC1: 'Reply on RC1', Erfan Goharian, 09 Aug 2023
The authors extend their sincere gratitude to the esteemed reviewers for dedicating their time and expertise to the assessment of this manuscript. This correspondence serves as a comprehensive account of the meticulous revisions and enhancements undertaken in accordance with their invaluable feedback. In thoughtful response to the insightful comments provided by the reviewers, several refinements have been incorporated into the manuscript. Notably, attached document contains additional elucidations aimed at elucidating and substantiating specific facets of the paper. We wish to reiterate our profound appreciation to the editor and the reviewers for their indispensable contributions, which have undeniably enriched the quality and scope of this work.
-
RC2: 'Comment on egusphere-2023-857', Anonymous Referee #2, 21 Jul 2023
This manuscript presents an original vision-based water level measurement approach applicable to real-time flood conditions. While I'm not able to evaluate many technical aspects relating to computer vision, AI and electronics involved in this technique, I found that the approach is rigorous, the methods are sound and openly available, and the research work is very well presented. I therefore feel that this contribution is significant for the water-related research community. Hereafter are a few comments for discussion and improvement of the paper.
It should be noted somewhere (e.g. in the abstract) that the system cannot work at night.
L142 The geographical coordinates of the site would be useful.
4.3 L266 Is the focal length fixed or variable?
Why don’t you estimate the DLT parameters directly? What is the advantage of estimating the intrinsic parameters beforehand?
Editorial:
Eq. 1: camera (camrea)
Citation: https://doi.org/10.5194/egusphere-2023-857-RC2 -
AC2: 'Reply on RC2', Erfan Goharian, 09 Aug 2023
The authors extend their sincere gratitude to the esteemed reviewers for dedicating their time and expertise to the assessment of this manuscript. This correspondence serves as a comprehensive account of the meticulous revisions and enhancements undertaken in accordance with their invaluable feedback. In thoughtful response to the insightful comments provided by the reviewers, several refinements have been incorporated into the manuscript. Notably, attached document contains additional elucidations aimed at elucidating and substantiating specific facets of the paper. We wish to reiterate our profound appreciation to the editor and the reviewers for their indispensable contributions, which have undeniably enriched the quality and scope of this work.
-
AC2: 'Reply on RC2', Erfan Goharian, 09 Aug 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 220 | 0 | 0 | 220 | 3 | 6 |
- HTML: 220
- PDF: 0
- XML: 0
- Total: 220
- BibTeX: 3
- EndNote: 6
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Seyed Mohammad Hassan Erfani
Corinne Smith
Zhenyao Wu
Elyas Asadi Shamsabadi
Farboud Khatami
Austin R. J. Downey
Jasim Imran
Erfan Goharian
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.