Next Article in Journal
Design and Experiment of Automatic Fish Bleeding Machine
Next Article in Special Issue
Hyperspectral Imaging for the Dynamic Mapping of Total Phenolic and Flavonoid Contents in Microgreens
Previous Article in Journal
AI-Driven Future Farming: Achieving Climate-Smart and Sustainable Agriculture
Previous Article in Special Issue
From Simulation to Field Validation: A Digital Twin-Driven Sim2real Transfer Approach for Strawberry Fruit Detection and Sizing
 
 
Article
Peer-Review Record

Random Reflectance: A New Hyperspectral Data Preprocessing Method for Improving the Accuracy of Machine Learning Algorithms

AgriEngineering 2025, 7(3), 90; https://doi.org/10.3390/agriengineering7030090
by Pavel A. Dmitriev *, Anastasiya A. Dmitrieva and Boris L. Kozlovsky
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
AgriEngineering 2025, 7(3), 90; https://doi.org/10.3390/agriengineering7030090
Submission received: 10 February 2025 / Revised: 10 March 2025 / Accepted: 18 March 2025 / Published: 20 March 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors
  • The abstract section needs improvement. It should clearly highlight the importance of the study, the objectives, the proposed approach, key outcomes, and future outlook
  • Lines (29-33): please add citations
  • Line (60): There are other promising approaches to improve classification accuracy. Please, mention me these promising approaches.
  • Lines (73-82): Rewrite and polish the content to clarify the main and sub-objectives, as they are unclear by the end of the introduction.
  • Object of study, rewrite
  • Add the coordinates of longitude and latitude for the study area.
  • Line (97): Cubert UHD-185 frame camera, add some information about this camera, including the manufacturer's name and its characteristics.
  • Line (103): Please complete it with a shortened sentence that explains this applied approach
  • The materials and methods section is too brief. Please provide more detailed information about the suggested approach, including algorithms like RF and GB, among others
  • It would be helpful to add a flowchart to explain the technology roadmap for this study.
  • The analysis of the study outcomes is too shallow. The authors should mention the results within the paragraphs when describing the outcomes.
  • The resolution of Figures 5 and 7 needs to be improved.
  • Lines (231, 235, 239, 242, and 249), the word 'method' is mentioned multiple times. Please avoid repeating the same word.
  • Please revise lines (255-260) for improved clarity and style.
  • Kindly add a section discussing the study's limitations and future directions.
  • Please, write the practical applications of your work in a separate section, before the conclusions and provide your good perspectives?.
  • The conclusion section needs to be rewritten completely, adhering to the established criteria for scientific writing. This section is currently quite weak.

This paper requires significant revisions before it can be considered complete. I would recommend a major revision. Thank you.

Author Response

Dear Reviewer!

Thank you so much for taking the time to review our manuscript! We appreciate your valuable feedback and constructive suggestions on our work. Your comments have made it much better. We retained revision marks on the revised manuscript in the ‘agriengineering-3496541 — Revised (tracked changes).docx’ file. The authors hope, the current version of the manuscript will meet your expectations regarding clarity and quality of presentation.
Special thanks for your constructive comments and recommendations for future research!

 

  • The abstract section needs improvement. It should clearly highlight the importance of the study, the objectives, the proposed approach, key outcomes, and future outlook

Response: Thank you for your comment! The abstract has been revised according to your comments. The changes can be tracked in a track-changed version.

 

  • Lines (29-33): please add citations

Response: Thank you for your comment! Citations were added.

 

  • Line (60): There are other promising approaches to improve classification accuracy. Please, mention me these promising approaches.

Response: Thank you for your comment! The paragraph has been corrected for clarity. Promising approaches are listed later in the text.

 

  • Lines (73-82): Rewrite and polish the content to clarify the main and sub-objectives, as they are unclear by the end of the introduction.

Response: Thank you for your comment! The paragraph has been corrected for clarity. The changes can be tracked in a track-changed version.

 

  • Object of study, rewrite

Response: Thank you for your comment! The subsection has been revised for clarity.

 

  • Add the coordinates of longitude and latitude for the study area.

Response: Thank you for your comment! Longitude and latitude coordinates of the study area were added to the manuscript. 47°13′ N; 39°39′ E.

 

  • Line (97): Cubert UHD-185 frame camera, add some information about this camera, including the manufacturer's name and its characteristics.

Response: Thank you for your comment! The camera manufacturer and country of manufacture have been added to the manuscript.

 

  • Line (103): Please complete it with a shortened sentence that explains this applied approach

Response: Thank you for your comment! We have added a sentence explaining this applied approach. The changes can be tracked in a track-changed version.

 

  • The materials and methods section is too brief. Please provide more detailed information about the suggested approach, including algorithms like RF and GB, among others

Response: Your commentary is appreciated! Pursuant to your feedback, the section entitled 'Materials and Methods' has been revised. With your permission, we have decided to refrain from describing such well-known algorithms as Random Forest and Gradient Boosting.

 

  • It would be helpful to add a flowchart to explain the technology roadmap for this study.

Response: Thank you for your suggestion! A flowchart explaining the technology roadmap for this study was added to the manuscript.

 

  • The analysis of the study outcomes is too shallow. The authors should mention the results within the paragraphs when describing the outcomes.

Response: Thank you for your comment! The Discussion section has been redesigned. The changes can be tracked in a track-changed version.

 

  • The resolution of Figures 5 and 7 needs to be improved.

Response: Thank you for your comment! The resolution of Figures 5 and 7 has been improved.

 

  • Lines (231, 235, 239, 242, and 249), the word 'method' is mentioned multiple times. Please avoid repeating the same word.

Response: Thank you for your comment! It has been attempted to eliminate the numerous repetitions of the word 'method'. The changes can be tracked in a track-changed version.

 

  • Please revise lines (255-260) for improved clarity and style.

Response: Thank you for your comment! The text has been revised. The changes can be tracked in a track-changed version.

 

  • Kindly add a section discussing the study's limitations and future directions.

Response: Thank you for your suggestion! Section ‘5. Limitations’ has been added to the manuscript.

 

  • Please, write the practical applications of your work in a separate section, before the conclusions and provide your good perspectives?.

Response: Thank you for your suggestion! Section ‘6. Future perspectives’ has been added to the manuscript.

 

  • The conclusion section needs to be rewritten completely, adhering to the established criteria for scientific writing. This section is currently quite weak.

Response: Thank you for your comment! The Conclusion section has been revised in accordance with your comments.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The paper presents a novel and engaging subject with a well-structured framework and clear presentation. However, several revisions are necessary before it can be considered for publication:

  1. Abstract: It should be rewritten to include numerical results and presented in a more quantitative manner.
  2. Lines 25–33: Please provide appropriate references to support the statements made in this section.
  3. Section 2.4: The authors should justify the decision to split the dataset into training and test subsets without incorporating a validation subset. While maintaining the current structure, the authors are encouraged to discuss the technical reasoning behind this choice, as it is seldom addressed in the literature. Additional insights can be found at: https://doi.org/10.1016/j.postharvbio.2025.113396.
  4. Section 3: The authors should elaborate on the superiority of the proposed model by discussing the structures and topologies of the machine learning models used.

Author Response

Dear Reviewer!

Thank you so much for taking the time to review our manuscript! We appreciate your valuable feedback and constructive suggestions on our work. Your comments have made it much better. We retained revision marks on the revised manuscript in the ‘agriengineering-3496541 — Revised (tracked changes).docx’ file. The authors hope, the current version of the manuscript will meet your expectations regarding clarity and quality of presentation.
Special thanks for your constructive comments and recommendations for future research!

 

The paper presents a novel and engaging subject with a well-structured framework and clear presentation. However, several revisions are necessary before it can be considered for publication:

Response: Thank you very much for taking the time to review this manuscript!

 

  1. Abstract: It should be rewritten to include numerical results and presented in a more quantitative manner.

Response: Thank you for your comment! The abstract has been revised according to your comments. The changes can be tracked in a track-changed version.

 

  1. Lines 25–33: Please provide appropriate references to support the statements made in this section.

Response: Thank you for your comment! Citations were added.

 

  1. Section 2.4: The authors should justify the decision to split the dataset into training and test subsets without incorporating a validation subset. While maintaining the current structure, the authors are encouraged to discuss the technical reasoning behind this choice, as it is seldom addressed in the literature. Additional insights can be found at: https://doi.org/10.1016/j.postharvbio.2025.113396.

Response: Thank you for your comment! A 5-fold cross-validation was used to adjust the hyperparameters and to assess efficacy. The changes can be tracked in a track-changed version.

 

  1. Section 3: The authors should elaborate on the superiority of the proposed model by discussing the structures and topologies of the machine learning models used.

Response: Thank you for your comment! The Discussion section has been redesigned. The changes can be tracked in a track-changed version.

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This article proposes a novel hyperspectral data preprocessing method named “Random Reflectance (RR),” which aims to improve the accuracy of machine learning algorithms in plant classification by generating synthetic spectral profiles. The study focuses on three species of maple trees (Acer campestre, A. negundo, and A. saccharinum) and demonstrates the effectiveness of this method in Random Forest (RF) and Gradient Boosting (GB) algorithms using hyperspectral image data acquired under laboratory conditions. Overall, the research has clear objectives, a well-designed methodology, and experimental results that show the RR method significantly improves classification accuracy, providing a new approach for hyperspectral data preprocessing.

However, there are some areas in the article that need improvement to facilitate readers' quick and accurate access to information. It is hoped that the authors will address these issues in subsequent revisions to further enhance the quality of the article.

 

1.The discussion of the limitations of existing technologies (such as noise sources and data redundancy) in the introduction is rather brief and lacks in-depth analysis of their specific impacts on practical applications. The authors should provide a more detailed discussion of the shortcomings of existing technologies, for example, how noise affects classification accuracy and how data redundancy increases computational costs. Additionally, more relevant literature should be cited to support these viewpoints.

 

2.The review of existing research is not comprehensive. The introduction provides a brief overview of existing hyperspectral data preprocessing methods, mentioning only some common techniques (such as Savitzky-Golay filtering and spectral smoothing) but omitting more complex or emerging methods. A more comprehensive review of existing studies, including advanced methods (e.g., preprocessing techniques based on deep learning) and emerging research directions, should be included. This would not only demonstrate the authors' deep understanding of the field but also provide stronger support for the novelty of the RR method.

 

3.The introduction cites a limited number of references, mainly focusing on the general applications of hyperspectral imaging technology, with a lack of in-depth citations related to specific preprocessing methods or relevant studies. More references, especially those related to hyperspectral data preprocessing, noise handling, and machine learning algorithms, should be included. This would enhance the persuasiveness of the introduction and provide readers with richer background information.

 

4.The logical structure is not clear enough. The transition from the introduction of hyperspectral imaging technology to the proposal of the RR method in the introduction is not smooth. The logical structure of the introduction should be optimized for greater coherence. For example, the content could be organized in the following order: (1) Background of hyperspectral imaging technology; (2) Limitations of existing technologies; (3) Proposal of the RR method and its objectives; (4) Potential contributions and application scenarios of the study.

 

5.Although the article mentions existing hyperspectral data preprocessing methods, it does not directly compare the RR method with other commonly used methods (such as Savitzky-Golay filtering and min-max normalization). This makes it difficult for readers to assess the relative advantages of the RR method in practical applications. The authors are advised to include comparative experiments with other preprocessing methods to demonstrate the unique advantages and applicability of the RR method in different datasets or classification tasks.

 

6.Although the study used hyperspectral data from four years, it did not deeply analyze the changes in spectral characteristics across different seasons or years and their impact on classification results. This may mask the special contributions of certain seasonal or annual data to classification outcomes. It is recommended to further analyze the dynamic changes in spectral characteristics across seasons and years and explore their impact on classification accuracy to reveal potential seasonal patterns.

 

7.During the review of the article, grammatical errors and overly long sentences may affect the readability and professionalism of the paper. The following is an analysis and suggestions for potential grammatical issues and long sentences in this article:

(1) “The use of these noise removal methods reduces the probability of false predictions, but slightly improves the accuracy of models, including those based on machine learning (ML) algorithms.”

Issue: The sentence structure is complex, and the logical relationship between the subject and predicate is not clear.

Suggestion: The authors should simplify the sentence structure and clarify the relationship between the subject and predicate.

(2) “The RR method was tested on Random Forest (RF) and Gradient Boosting (GB) algorithms in the task of classifying three maple species based on time series of their spectral characteristics over four years of research.”

Issue: The tenses of “was tested” and “classifying” are inconsistent.

Suggestion: The authors should make revisions to maintain consistent tenses.

(3) The sentence is too long. “The prospects of using synthetic SP for maple classification are confirmed by the results of principal component analysis (PCA) (Figure 3). Synthetic SP are much more compact in the plane of the first two principal components than the original SP. Maple species are well separated from each other (especially A. negundo).”

Issue: The sentence is lengthy and contains too much information, making it difficult to read.

Suggestion: The authors should break the sentence into several shorter ones to enhance readability.

(4) Too many clauses. “The following trend was observed – the lower the classification accuracy based on real spectral profiles, the greater the positive effect of using synthetic spectral profiles.”

Issue: The sentence structure is complex, containing multiple clauses, which affects readability.

Suggestion: Simplify the sentence structure and use more direct expressions.

There are some issues in terms of grammar and sentence structure, such as complex sentence structures and overly long sentences. These issues may affect the readability and professionalism of the article. The authors are advised to review the entire text and make revisions in these areas to improve the overall quality of the article.

 

8.In scientific papers, the “Conclusions” section is a crucial part that summarizes research findings, emphasizes contributions, and proposes future work directions. However, the analysis of defects and shortcomings in the “Conclusions” section (5. Conclusions) is too brief and lacks a comprehensive discussion of research contributions, limitations, and practical applications. It is recommended that the authors provide a more comprehensive summary of the main findings in the conclusion, highlight the novelty of the method, discuss limitations, and propose directions for future work to enhance the completeness and depth of the conclusions.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

There are some issues in terms of grammar and sentence structure, including complex sentence structures and overly long sentences. These issues may affect the readability and professionalism of the article. The author is advised to review the entire text and make revisions in these areas to improve the overall quality of the article.

Author Response

Dear Reviewer!

Thank you so much for taking the time to review our manuscript! We appreciate your valuable feedback and constructive suggestions on our work. Your comments have made it much better. We retained revision marks on the revised manuscript in the ‘agriengineering-3496541 — Revised (tracked changes).docx’ file. The authors hope, the current version of the manuscript will meet your expectations regarding clarity and quality of presentation.
Special thanks for your constructive comments and recommendations for future research!

 

This article proposes a novel hyperspectral data preprocessing method named “Random Reflectance (RR),” which aims to improve the accuracy of machine learning algorithms in plant classification by generating synthetic spectral profiles. The study focuses on three species of maple trees (Acer campestre, A. negundo, and A. saccharinum) and demonstrates the effectiveness of this method in Random Forest (RF) and Gradient Boosting (GB) algorithms using hyperspectral image data acquired under laboratory conditions. Overall, the research has clear objectives, a well-designed methodology, and experimental results that show the RR method significantly improves classification accuracy, providing a new approach for hyperspectral data preprocessing.

However, there are some areas in the article that need improvement to facilitate readers' quick and accurate access to information. It is hoped that the authors will address these issues in subsequent revisions to further enhance the quality of the article.

Response: Thank you very much for taking the time to review this manuscript. We agree with these suggestions and have revised the manuscript accordingly. Our responses are provided below.

 

1.The discussion of the limitations of existing technologies (such as noise sources and data redundancy) in the introduction is rather brief and lacks in-depth analysis of their specific impacts on practical applications. The authors should provide a more detailed discussion of the shortcomings of existing technologies, for example, how noise affects classification accuracy and how data redundancy increases computational costs. Additionally, more relevant literature should be cited to support these viewpoints.

Response: Thank you for your comment! The abstract has been revised according to your comments. The changes can be tracked in a track-changed version.

 

2.The review of existing research is not comprehensive. The introduction provides a brief overview of existing hyperspectral data preprocessing methods, mentioning only some common techniques (such as Savitzky-Golay filtering and spectral smoothing) but omitting more complex or emerging methods. A more comprehensive review of existing studies, including advanced methods (e.g., preprocessing techniques based on deep learning) and emerging research directions, should be included. This would not only demonstrate the authors' deep understanding of the field but also provide stronger support for the novelty of the RR method.

Response: Thank you for your comment! We have given examples of such methods in the Introduction section (lines: 56-59).

 

3.The introduction cites a limited number of references, mainly focusing on the general applications of hyperspectral imaging technology, with a lack of in-depth citations related to specific preprocessing methods or relevant studies. More references, especially those related to hyperspectral data preprocessing, noise handling, and machine learning algorithms, should be included. This would enhance the persuasiveness of the introduction and provide readers with richer background information.

Response: Thank you for your comment! References related to hyperspectral data preprocessing have been added to the Introduction section.

 

4.The logical structure is not clear enough. The transition from the introduction of hyperspectral imaging technology to the proposal of the RR method in the introduction is not smooth. The logical structure of the introduction should be optimized for greater coherence. For example, the content could be organized in the following order: (1) Background of hyperspectral imaging technology; (2) Limitations of existing technologies; (3) Proposal of the RR method and its objectives; (4) Potential contributions and application scenarios of the study.

Response: Thank you for your comment! We have endeavoured to redesign the Introduction section in line with your comments.

 

5.Although the article mentions existing hyperspectral data preprocessing methods, it does not directly compare the RR method with other commonly used methods (such as Savitzky-Golay filtering and min-max normalization). This makes it difficult for readers to assess the relative advantages of the RR method in practical applications. The authors are advised to include comparative experiments with other preprocessing methods to demonstrate the unique advantages and applicability of the RR method in different datasets or classification tasks.

Response: Thanks for your suggestions! We compared the performance of the RR method with min-max normalisation and PCA. The changes can be tracked in a track-changed version.

 

6.Although the study used hyperspectral data from four years, it did not deeply analyze the changes in spectral characteristics across different seasons or years and their impact on classification results. This may mask the special contributions of certain seasonal or annual data to classification outcomes. It is recommended to further analyze the dynamic changes in spectral characteristics across seasons and years and explore their impact on classification accuracy to reveal potential seasonal patterns.

Response: Thank you for your comment! In this manuscript we would like to focus on presenting a new method of hyperspectral data preprocessing for species classification tasks. Therefore, with your permission, we would not like to consider this issue in this paper. Although it is very interesting and topical. Despite the large number of publications on this topic, no clear answer to the relationship between season and classification accuracy has been established yet. In general, consideration of this issue deserves a separate article. Based on the large factual material we plan to do it in the future.

 

7.During the review of the article, grammatical errors and overly long sentences may affect the readability and professionalism of the paper. The following is an analysis and suggestions for potential grammatical issues and long sentences in this article:

(1) “The use of these noise removal methods reduces the probability of false predictions, but slightly improves the accuracy of models, including those based on machine learning (ML) algorithms.”

Issue: The sentence structure is complex, and the logical relationship between the subject and predicate is not clear.

Suggestion: The authors should simplify the sentence structure and clarify the relationship between the subject and predicate.

(2) “The RR method was tested on Random Forest (RF) and Gradient Boosting (GB) algorithms in the task of classifying three maple species based on time series of their spectral characteristics over four years of research.”

Issue: The tenses of “was tested” and “classifying” are inconsistent.

Suggestion: The authors should make revisions to maintain consistent tenses.

(3) The sentence is too long. “The prospects of using synthetic SP for maple classification are confirmed by the results of principal component analysis (PCA) (Figure 3). Synthetic SP are much more compact in the plane of the first two principal components than the original SP. Maple species are well separated from each other (especially A. negundo).”

Issue: The sentence is lengthy and contains too much information, making it difficult to read.

Suggestion: The authors should break the sentence into several shorter ones to enhance readability.

(4) Too many clauses. “The following trend was observed – the lower the classification accuracy based on real spectral profiles, the greater the positive effect of using synthetic spectral profiles.”

Issue: The sentence structure is complex, containing multiple clauses, which affects readability.

Suggestion: Simplify the sentence structure and use more direct expressions.

There are some issues in terms of grammar and sentence structure, such as complex sentence structures and overly long sentences. These issues may affect the readability and professionalism of the article. The authors are advised to review the entire text and make revisions in these areas to improve the overall quality of the article.

Response: We would like to express our gratitude for your meticulous and comprehensive evaluation of the manuscript. We have endeavoured to incorporate all of your observations into the document. It is our sincere hope that this revised version will meet your expectations. The changes can be tracked in a track-changed version.

 

8.In scientific papers, the “Conclusions” section is a crucial part that summarizes research findings, emphasizes contributions, and proposes future work directions. However, the analysis of defects and shortcomings in the “Conclusions” section (5. Conclusions) is too brief and lacks a comprehensive discussion of research contributions, limitations, and practical applications. It is recommended that the authors provide a more comprehensive summary of the main findings in the conclusion, highlight the novelty of the method, discuss limitations, and propose directions for future work to enhance the completeness and depth of the conclusions.

Response: Thank you for your comment! The Conclusions section has been revised.

 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors improved the manuscript but the discussion section is still needed to improve for covering results section.

Author Response

  • The authors improved the manuscript but the discussion section is still needed to improve for covering results section.

Response: Thank you for your comment! The Discussion section has been revised. We have tried to take your comments into account. We hope we have succeeded. We would like to express our gratitude once again for the time you have invested in reviewing our manuscript! The changes can be tracked in a track-changed version.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

After meticulous revision and optimization, the article has witnessed a significant improvement in terms of the depth of content, coherence of logic, and precision of language.

Comments on the Quality of English Language

The article needs to be fully checked and necessary grammatical corrections are made to improve its quality. No obvious errors are made, so that it can fully meet the publication requirements.

Author Response

After meticulous revision and optimization, the article has witnessed a significant improvement in terms of the depth of content, coherence of logic, and precision of language.

Response: Dear Reviewer, Your contribution to the enhancement of the manuscript is greatly appreciated!

 

The article needs to be fully checked and necessary grammatical corrections are made to improve its quality. No obvious errors are made, so that it can fully meet the publication requirements.

Response: Thank you for your comment! The manuscript was subjected to a thorough revision to identify and rectify any grammatical errors. The changes can be tracked in a track-changed version.

 

Author Response File: Author Response.docx

Back to TopTop