River Ice Regime Recognition Based on Deep Learning: Ice Concentration, Area, and Velocity
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis study aims to address the challenging task of recognizing river ice density, area, velocity, and melting stage based on camera imagery. The topic of the paper is of significant importance, particularly in developing the real-time short-term forecasts of ice floods. However, this manuscript falls below the publication standards of this journal, and major revisions are necessary to improve the clarity and readability of the paper. Specific comments are listed below.
The paper could be more concise and better structured to provide a clearer understanding of the objectives and research questions of the paper.
For instance, the method section presents the results and discussion related to the models (such as Table 2), while some information that could be included in the method section is found in the introduction (specifically, information in lines 82-120).
Figure 5 needs revision to provide a better understanding of the study.A detailed description of the dataset preprocessing is essential. In the process of preparing images for analysis, it is critical to discuss the various steps involved, such as data preprocessing, feature extraction, and normalization. Data preprocessing may include tasks such as resizing, cropping, and augmenting images to ensure compatibility with the deep learning models being used.Have the hyperparameters been optimized in this study to enhance the performance of the deep learning models? Optimizing hyperparameters can significantly impact the convergence and generalization of the models, ultimately influencing their accuracy and loss on test data.In Figure 16, it is important to present the average accuracy and loss for the test data to provide insights into the performance of the deep learning model for unseen data. Additionally, it is essential to present the accuracy and loss of other models (for train, validation, and test data) to facilitate a comparative analysis of their performance.The detection of early flood warnings is a critical aspect that should be thoroughly addressed in the paper. Providing uncertainty quantification can aid decision-makers in understanding the uncertainty associated with warning detection and taking appropriate action. Critical discussion on comparitive analysis of different studies should be presented. What is the objective of presenting Figure 17, while all images predict the ice drifting stage. The keywords should be revised to better describe the objectives of the paper. A minimum of 5 keywords is recommended.The grammar should be carefully reviewed. In academic writing, past and passive sentences are typically preferred. Place line 143 after the image labeled as Figure 3.
In conclusion, while the study demonstrates promising work in the field of image and video processing using deep learning, there is a need for more detailed and comprehensive descriptions of the modeling processes and architectures.
Comments on the Quality of English Language
The grammar should be carefully reviewed. In academic writing, past and passive sentences are typically preferred.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsYang proposed a river ice regime recognition based on deep learning. Overall, the results and the novelty of the manuscript is quite good. However, the manuscript must be improved before accepted.
1. The manuscript should improve English and have been check by professional service. I spoil several grammar error and typo in the manuscript.
2. The manuscript should be re-written in concise and clearly.
3. The main contribution as well as the significance of the research should be clearly shown in abstract and introduction.
4. In introduction, the shortcoming of other research, and the survey on the state-of-art methodology should be made.
5. Improve the size and the quality of Fig. 3 and Fig 4.
6. The quality and the way you present in figure 5 is lowed. Please refer to the researches to know how to present Process Overview : (A flexible, and wireless LED therapy patch for skin wound photomedicine with IoT-connected healthcare application, A smart LED therapy device with an automatic facial acne vulgaris diagnosis based on deep learning and internet of things application, Smart Low Level Laser Therapy System for Automatic Facial Dermatological Disorder Diagnosis)
7. The figure 10 don’t show how the neural network works, it is quite overall. Please refer to the research to how to present the Feedforward Neural Network. (Enhanced precision of real-time control photothermal therapy using cost-effective infrared sensor array and artificial neural network)
Comments on the Quality of English LanguageThe manuscript should improve English and have been check by professional service. I spoil several grammar error and typo in the manuscript.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsAccept in present form
Author Response
We feel great thanks for your professional review work on our article.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors satisfied my revision. I recommend the manuscript for publication.
Author Response
We feel great thanks for your professional review work on our article.
