Microscopic Insect Pest Detection in Tea Plantations: Improved YOLOv8 Model Based on Deep Learning

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript titled " Microscopic insect pest detection in tea plantations: Improved YOLOv8 model based on deep learning" presents research that develops an improved YOLOv8 network model based on deep learning, providing an efficient and accurate technical means for intelligent detection and management of tea pests. There are some revisions to improve the quality of this manuscript.

1.     The line numbering is missing.

2.     The last paragraph of the introduction is a bit verbose, please simplify it.

3.     The authors should add information (company, city, country) about the materials used in this manuscript, including s iPhone 14 Pro Max and Redmi K50 etc.

4.     Why only training and test sets were used without validation sets?

5.     The self-evident nature of the chart needs to be strengthened. Supplementary explanation of each sub-graph information is provided as shown in Figure 1. The resolution of the picture needs to be improved, and some fonts are too small.

6.     It is recommended to add a discussion section. An in-depth explanation of the research results and comparison with current relevant research should be added to the discussion.

7.     The language is not simple enough, so it is recommended to make overall changes.

8.     The conclusion section summarizes the research innovations and main results.

Comments on the Quality of English Language

The language is not simple enough, so it is recommended to make overall changes.

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) The line numbering is missing.

Modification instructions: Thank you for your valuable feedback. We have added continuous line numbers to the manuscript as per the template requirements, which will facilitate your and the editorial department's reference to the information within the text. We sincerely appreciate your meticulous review and look forward to your further guidance.

 

(2) The last paragraph of the introduction is a bit verbose, please simplify it.

Modification instructions: Thank you for your valuable feedback. We have simplified the last paragraph of the introduction to enhance the clarity and professionalism of the article. The revised content more directly states the core objectives and contributions of this study, while reducing verbose expressions.

The revised final paragraph of the introduction is as follows:

This study proposes an improved YOLOv8 network model for the detection of pests in tea gardens, which is challenging to identify with the naked eye. The model is enhanced to improve the detection speed and accuracy of tiny pests in tea leaves. To accelerate model convergence, simplify the computation process, and accurately locate target positions, the original loss function is replaced with the SIoU (Smoothed Intersection over Union). Additionally, the AKConv (Attention with Kernel Convolutions) is employed to enhance the accuracy of target detection, reduce model parameters, and computational overhead. Meanwhile, BiFormer (Bi-Level Routing Attention Vision Transformer) is embedded to strengthen the recognition efficiency for targets with partial shape damage, making the model's computational allocation and target perception more flexible. The method proposed in this study provides a practical research approach and important reference for solving the problem of microscopic pest identification in tea leaves, offering an effective way to ensure the high-quality development and healthy growth of the tea industry in Yunnan.

 

(3) The authors should add information (company, city, country) about the materials used in this manuscript, including s iPhone 14 Pro Max and Redmi K50 etc.

Modification instructions: Thank you for your suggestions. We have included detailed information about the materials used in the Materials and Methods section of our paper, including the company, city, and country. We sincerely appreciate your thorough review and valuable input.

 

(4) Why only training and test sets were used without validation sets?

Modification instructions: Thank you very much for your feedback. To ensure the objectivity and effectiveness of the ablation experiment results, we used a validation set. Due to our oversight, this was not indicated in the text, and we have made the necessary corrections.

 

(5) The self-evident nature of the chart needs to be strengthened. Supplementary explanation of each sub-graph information is provided as shown in Figure 1. The resolution of the picture needs to be improved, and some fonts are too small.

Modification instructions: Thank you for your attention to the clarity of the figures in our paper. Following your suggestions, we have made the following improvements to the figures:

1. We have added supplementary explanations to each subplot to ensure that the information in each figure is self-explanatory, allowing readers to easily understand the content of the figures.
2. We have increased the resolution of the figures to ensure that they remain clear when viewed on different display devices.
3. We have adjusted the font size in the figures to ensure that all text is clearly legible and does not pose any reading barriers for the readers.

      We believe that these modifications will make the figures more intuitive and easy to understand, contributing to the enhancement of the quality of the paper. We greatly appreciate your valuable feedback and look forward to your further guidance.

 

(6) It is recommended to add a discussion section. An in-depth explanation of the research results and comparison with current relevant research should be added to the discussion.

Modification instructions: Thank you for your valuable feedback. Following your suggestion, we have added a detailed discussion section in the paper. In this section, we have provided an in-depth explanation of the research findings and compared them with current relevant studies.

The added discussion section is as follows:

4. Discussion

     Pest infestations in tea gardens are one of the common issues in the process of tea cultivation. To achieve rapid and accurate identification of early-stage minor pests in tea gardens, this study proposes an improved YOLOv8 network model for tea pest detection, addressing issues such as small datasets and difficulty in extracting target pest phenotypic features in the tea pest detection process. The model significantly enhances detection accuracy and robustness by introducing advanced technologies such as the SIoU loss function, AKConv, and BiFormer. Experimental results show that the improved model has achieved a precision rate of 98.16% in tea pest detection tasks, which is a 2.62% increase compared to the original YOLOv8 network, demonstrating its efficiency and accuracy in tea pest detection. This improvement significantly enhances the model's ability to recognize minor pests, which is of great significance for ensuring the healthy development of the tea industry, especially in the context of Yunnan's ecological tea industry.

      The improved YOLOv8 network model in this study not only enhances the model's learning ability for small target pest samples in tea gardens but also improves the acquisition capability of target location information, thereby enhancing the model's perceptual performance for targets. Compared with existing studies, it is noted that Solimani et al. [8] proposed a lightweight YOLOv8n-ShuffleNetv2-Ghost-SE model, which achieved an average precision of 91.4% and a recall rate of 82.6%, while this study has improved the average precision and recall rate by 6.76% and 15.45% respectively. Fuentes et al. [18] proposed a tomato pests and disease detection network that combines VGG and ResNet as deep feature extractors. However, the network model detected an average precision of only 85.98%, which is 12.18% lower than that of this study, indicating a higher detection accuracy in this research. Dai et al. [19] introduced the Swin Transformer mechanism and improved feature fusion strategy into the YOLOv5m model, achieving a recall rate of 93.1%, an F1 score of 94.38%, and an average precision of 96.4%, but this study has significantly improved by 4.95%, 3.07%, and 1.76% respectively. The improved YOLOv8 network model in this study has shown stronger performance in feature extraction and target localization for tea pest detection, achieving higher detection accuracy through specially optimized loss functions and network structures. In addition, by introducing AKConv and BiFormer, the model has shown stronger performance in feature extraction and target localization, which are less involved in existing studies.

      Although this study has performed well in the identification and detection of specific foreign targets such as minor tea pests, there are still certain limitations. First, the model's performance has mainly been tested under specific environmental conditions in Yunnan tea gardens, and its performance in other regions and different environmental conditions has not been fully verified, so there is a limitation in environmental adaptability. Secondly, the main types of pests studied are relatively limited, and the model's recognition ability and accuracy for a wider range of pest types still need further testing and optimization. In addition, due to the limited scale of the training dataset, this may limit the model's generalization ability in recognizing new types of pests. To enhance the model's generalization, it is recommended to conduct more in-depth data augmentation processing on the dataset in this study. Specifically, the diversity of the dataset can be enriched by using techniques such as generative adversarial networks, feature fusion, oversampling, and undersampling, thereby enhancing the model's recognition ability for different pest types. Based on this, the model's performance may be affected by environmental changes, the diversity of pest types, and dataset biases. Therefore, potential future directions can be developed from the above content to further promote the development of tea pest detection technology and provide scientific and technological support for the sustainable development of the tea industry.

 

(7) The language is not simple enough, so it is recommended to make overall changes.

Modification instructions: Thank you for your attention to the conciseness of the language in our paper. We have thoroughly revised the entire manuscript to ensure that the language is more concise and clear. We paid particular attention to the following points:

1. We have simplified complex sentence structures and adopted more direct expressions.
2. We have streamlined lengthy sentences and paragraphs, removing unnecessary repetitions and verbosity.
3. We have optimized the use of technical terminology, ensuring accuracy and consistency in terminology.
4. We have enhanced the logic of the paper, making the connections between arguments and evidence clearer.

 

(8) The conclusion section summarizes the research innovations and main results.

Modification instructions: Thank you for your valuable comments on the conclusion section of this study. Following your suggestions, we have further refined and summarized the conclusion section.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript (agriculture-3194117) introduces an improved YOLOv8 network model for detecting micropests in tea gardens, enhancing feature extraction and accuracy through modifications such as the SIoU loss function, AKConv, and Vision Transformer with bilevel routing attention. The model achieves a detection accuracy of 98.16%, outperforming the YOLOv7, Faster RCNN, and SSD models, contributing to the efficient identification of early-stage pests in tea cultivation.

The manuscript is interesting, but significant revisions are needed. Keywords need to be organised alphabetically. In the introduction, a clearer relationship regarding the importance of "tea-producing regions" should be provided. Additionally, a better description of "intelligent management of pest control" should be presented, alongside a discussion of the importance of digital agriculture, machine learning classification models, and artificial intelligence, specifically how they are effectively applied. An introductory approach on these topics should be included in the introduction section. In the final paragraph of the introduction, the hypotheses and objectives of the study need to be adequately described, as they are somewhat confusing.

All figure legends need to be expanded and clarified. For example, all elements that appear, such as colours, sample numbers, and regions, should be clearly identified. The reader should not have to refer back to the text or attempt to infer what the authors are trying to convey in the figure. The legends must be self-explanatory. Improve all of them – this is mandatory!

The materials and methods section, which includes many figures that represent results, needs to be restructured. For instance, figures and tables that describe results should appear in the results section, not in the materials and methods section. Furthermore, this section should be titled "3. Results".

Figure 14 is of poor quality and difficult to visualise. A general comment on the figures: they need to be improved in terms of quality. Additionally, the fonts must be enlarged, as they are too small and hard to read, and the lines should be slightly thicker. Are all the figures necessary in the manuscript? There is considerable repetition and redundant information. For example, some could be moved to a supplementary file, indicated in the text as (Figure Sx).

In the results section, only results should be described without any opinions or discussion of the data. The manuscript currently lacks a discussion section, which is problematic because, given the amount of data, the authors could reinforce and deepen our understanding of the topic. The absence of a discussion section leaves the data vague, disjointed, and without perspective. Even in the results section, where there are few discussion passages, there is no theoretical or bibliographical support. In this regard, the manuscript is inadequate in my view.

The conclusions section needs to establish the study's limitations, future perspectives, and potential new directions that could be explored with the methods and technology presented and applied by the authors.

The references section must be improved with articles from the last five years at least. 

Comments on the Quality of English Language

The language should be reviewed for clarity, reducing verbosity and fluency errors.

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) The manuscript is interesting, but significant revisions are needed. Keywords need to be organised alphabetically. In the introduction, a clearer relationship regarding the importance of "tea-producing regions" should be provided. Additionally, a better description of "intelligent management of pest control" should be presented, alongside a discussion of the importance of digital agriculture, machine learning classification models, and artificial intelligence, specifically how they are effectively applied. An introductory approach on these topics should be included in the introduction section. In the final paragraph of the introduction, the hypotheses and objectives of the study need to be adequately described, as they are somewhat confusing.

Modification instructions: Thank you for your positive feedback and valuable suggestions on our paper. Based on your comments, we have made the following significant revisions to the manuscript:

1. We have reorganized the keywords in alphabetical order to improve readability and search efficiency.
2. In the introduction, we have added a clear description of the importance of 'tea-producing regions' and provided relevant background information. At the same time, we have elaborated on 'intelligent pest management' in more detail and discussed the application and significance of digital agriculture, machine learning classification models, and artificial intelligence in tea pest management.
3. In the last paragraph of the introduction, we have rephrased the research hypotheses and objectives to ensure they are more explicit and understandable.

     We believe these modifications will make the content of the paper clearer and more persuasive. We greatly appreciate your suggestions and look forward to your further guidance.

 

(2) All figure legends need to be expanded and clarified. For example, all elements that appear, such as colours, sample numbers, and regions, should be clearly identified. The reader should not have to refer back to the text or attempt to infer what the authors are trying to convey in the figure. The legends must be self-explanatory. Improve all of them – this is mandatory!

Modification instructions: Thank you for your review and valuable feedback on the legends of the figures in our paper. We understand that the clarity and self-explanatory nature of the legends are crucial for readers to understand the content of the figures. Based on your suggestions, we have made the following improvements to the legends of all figures:

1. Expanded and clarified legends: We have detailed and clarified the legends for all figures to ensure that each legend clearly explains every element in the figure, including colors, sample sizes, different regions, etc.
2. Clear identification of elements: We have ensured that all elements appearing in the figures are clearly identified and explained, allowing readers to understand the information conveyed by the figures without referring to the main text.
3. Enhanced self-explanatory nature: We have rewritten the legends to make them more self-explanatory, so that readers can obtain all necessary information directly from the legends.
4. Consistency and accuracy: We have checked the consistency and accuracy of the information in the legends with the content of the figures, ensuring that they match perfectly and that all terms and abbreviations are defined in the legends.

 

(3) The materials and methods section, which includes many figures that represent results, needs to be restructured. For instance, figures and tables that describe results should appear in the results section, not in the materials and methods section. Furthermore, this section should be titled "3. Results".

Modification instructions: Thank you for your valuable comments on this research. Following your suggestions, we have adjusted the structure of the paper, particularly reorganizing the Materials and Methods section. Now, all figures and tables that describe the results have been moved from the Materials and Methods section to the Results section, and we have ensured that these figures and tables match the narrative of the results. We believe these modifications will make the structure of the paper clearer and help readers better understand the research findings. We sincerely appreciate your meticulous review and professional advice, and we look forward to your further feedback.

 

(4) Figure 14 is of poor quality and difficult to visualise. A general comment on the figures: they need to be improved in terms of quality. Additionally, the fonts must be enlarged, as they are too small and hard to read, and the lines should be slightly thicker. Are all the figures necessary in the manuscript? There is considerable repetition and redundant information. For example, some could be moved to a supplementary file, indicated in the text as (Figure Sx).

Modification instructions: Thank you for your valuable feedback on the quality of the figures in our paper. We have carefully considered your suggestions and have made the following improvements to the figures in the paper:

1. Enhanced figure quality: We have redone the figures with higher resolution and clearer layouts to ensure the quality of the images and the effectiveness of the visualization.
2. Font size adjustment: We have appropriately increased the font size in all figures to ensure they are clear and legible.
3. Line thickness optimization: The lines in the figures have been thickened to improve the readability and aesthetics of the charts.
4. Assessment of figure necessity: We have conducted a thorough review of all figures to ensure that each one makes a substantial contribution to the content of the paper. We have removed some figures that were redundant or overly repetitive.

 

(5) In the results section, only results should be described without any opinions or discussion of the data. The manuscript currently lacks a discussion section, which is problematic because, given the amount of data, the authors could reinforce and deepen our understanding of the topic. The absence of a discussion section leaves the data vague, disjointed, and without perspective. Even in the results section, where there are few discussion passages, there is no theoretical or bibliographical support. In this regard, the manuscript is inadequate in my view.

Modification instructions: Thank you for your valuable comments on this research. We have carefully considered your suggestions and have made corresponding revisions and additions to the paper. Here are the responses to your specific questions:

1. Regarding the addition of the discussion section, we have included a detailed discussion in the paper, where we have conducted an in-depth analysis of the research results and compared them with existing relevant studies. We have thoroughly discussed the application of the improved YOLOv8 network model in tea pest detection, as well as its advantages in feature extraction and target localization. Furthermore, we have discussed the limitations of the model and future research directions, with the aim of providing technological support for the sustainable development of the tea industry.
2. For the results section, we have ensured that all data and results are presented in an objective and clear manner, avoiding the inclusion of any subjective opinions or discussions. We have made sure that the results section contains only a direct description of the experimental results, without interpreting or inferring the data.
3. We have also summarized the main findings of the research in the conclusion section and proposed the challenges the model may face in practical applications and future directions for improvement. We believe that these additions and modifications will help readers to understand our research more comprehensively and provide valuable references for subsequent studies.

 

(6) The conclusions section needs to establish the study's limitations, future perspectives, and potential new directions that could be explored with the methods and technology presented and applied by the authors.

Modification instructions: Thank you for your valuable comments on this research. We have carefully considered your suggestions and have made corresponding revisions and additions to the conclusion section of the paper. Here are the responses to your specific questions:

1. We have clearly stated the limitations of this study in the conclusion section, including the testing limitations of the model under specific environmental conditions, the limited focus on pest species, and the constraints of the training dataset size. We recognize that these limitations may affect the model's generalizability and adaptability, and we have discussed them in detail in the text.
2. Regarding future research directions, we have proposed several potential new avenues, including testing the model across regions and under multi-environmental conditions, increasing the variety of pest samples to improve recognition capabilities, exploring more efficient model optimization strategies, and deploying the model into actual tea garden monitoring systems to achieve real-time monitoring and management of tea pests.
3. We believe that with these modifications, our paper is not only more complete in structure but also more in-depth and comprehensive in content. We look forward to these improvements enhancing the academic value of the paper and providing more valuable references for research in the field of tea pest detection.

      Thank you again for your suggestions and support.

 

(7) The references section must be improved with articles from the last five years at least.

Modification instructions: Thank you for your valuable feedback, and we recognize the importance of incorporating the latest research findings in the references section of our paper. Following your suggestions, we have made the following improvements to the references section:

1. Update of recent literature: We have expanded the list of references to ensure that it includes relevant research articles published within the last five years, reflecting the latest developments in the field.
2. Relevance assessment: We have carefully selected the additional literature to ensure that they are highly relevant to our research topic and can provide support for the research background and discussion in the paper.
3. Citation integrity: We have checked the completeness and accuracy of all citations to ensure that each citation complies with the journal's citation format requirements.
4. Correspondence in the text: We have also appropriately cited the latest literature in the main body of the paper to strengthen the arguments and evidence support of the thesis.

Reviewer 3 Report

Comments and Suggestions for Authors

The paper presents an innovative approach to detecting microscopic pests in tea plantations using an improved YOLOv8 model. The authors focus on addressing the challenges of small datasets and difficulties in feature extraction in tea pest detection. The study enhances the YOLOv8 framework by incorporating a new loss function (SIoU), AKConv for feature extraction, and the BiFormer module to improve attention mechanisms. The improvements result in significant performance gains in detection accuracy.

What must be improved: 

1. In the title there is word "microscopic insect pests". I didnt found anywhere in the text eplanation what does it mean. It is litle confusing...  Are these pests only visible under a microscope, or does "microscopic" refer to their size relative to normal pests? So please explaine.

2. Paragraph 2.2. The process of data enhancement could benefit from more explanation. Clarifying in more detail how does random perturbation work? Its not quiet cleare to me.

3. Paragraph  2.2: It would be helpful to explain why these particular augmentation techniques were chosen (brightness adjustment, contrast adjustment, etc.) and whether any others were considered or tested.

4. Paragraph 2.3.2 I think that authors should explain more explicitly how the AKConv structure adapts dynamically and what advantages it provides over conventional convolution operations.

5. Paragraph 3.4.2: Authors must include more details about how the external validation was conducted. Were there significant differences between the external validation dataset and the original dataset in terms of pest appearance, lighting, or other environmental conditions?

General comment: Improve figure captions and explanations for mathematical equations.

All other comments you can find in pdf document.

Comments for author File: Comments.pdf

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) In the title there is word "microscopic insect pests". I didnt found anywhere in the text eplanation what does it mean. It is litle confusing...  Are these pests only visible under a microscope, or does "microscopic" refer to their size relative to normal pests? So please explaine.

Modification instructions: Thank you for your meticulous review and valuable feedback. In response to the questions you raised, we have made professional revisions to the relevant parts of the paper. The term 'microscopic insect pests' in this study refers to insects that are smaller in size compared to common pests. These tiny pests are extremely small in stature and typically require the use of magnification tools (such as the macro lens used in this study) in conjunction with image capture devices (smartphones) to be observed. Therefore, the improved model proposed in this study aims to enhance the detection efficiency and accuracy of these tiny pests, in order to achieve more effective pest management. We appreciate your suggestions and look forward to your further guidance.

 

(2) Paragraph 2.2. The process of data enhancement could benefit from more explanation. Clarifying in more detail how does random perturbation work? Its not quiet cleare to me.

Modification instructions: Thank you for your valuable comments on this research. In response to your questions about the data augmentation process, we have made detailed additions and revisions to Section 2.2 of the paper to more clearly explain the working mechanism of random perturbations.

       During the data augmentation process, random perturbation is a technique that simulates new data by applying random variations to the original image data. Specifically, random perturbations increase the diversity of the dataset and the generalization ability of the model by introducing random noise at the pixel level of the image or by performing random transformations on the geometric structure of the image (such as rotation, scaling, shearing, etc.). This method can effectively simulate the various changes that pest images may exhibit in natural environments, thereby improving the model's ability to recognize pests under different postures and lighting conditions.

      In this study, we have employed the following random perturbation techniques:

1. Random geometric transformations: Randomly rotate, scale, and translate images to simulate pests in different positions and orientations within the image.
2. Random color adjustments: Enhance the model's adaptability to different color variations by randomly altering the brightness, contrast, and saturation of the images.

     Through the detailed improvements in the revised manuscript, we believe we can better meet the reviewers' requirements and enhance the quality and readability of the paper.

 

(3) Paragraph  2.2: It would be helpful to explain why these particular augmentation techniques were chosen (brightness adjustment, contrast adjustment, etc.) and whether any others were considered or tested.

Modification instructions: Thank you for your valuable feedback on our paper. Following your suggestions, we have added a detailed explanation of the selected image enhancement techniques in Section 2.2 of the paper. We chose brightness adjustment, contrast enhancement, and random deflection processing because these techniques can effectively simulate different lighting conditions and enhance image details, which is particularly important for pest image recognition. We also considered other techniques, such as histogram equalization and gamma correction, but found in preliminary tests that the selected techniques were more effective in improving model performance. Therefore, we decided to focus on these techniques. All revisions have been highlighted in red in the revised manuscript for easy identification. We appreciate your suggestions once again and hope that these modifications meet your requirements.

 

(4) Paragraph 2.3.2 I think that authors should explain more explicitly how the AKConv structure adapts dynamically and what advantages it provides over conventional convolution operations.

Modification instructions: Thank you for your valuable feedback on the paper. Following your suggestions, we have provided a more detailed discussion on the dynamic adaptability of the AKConv structure and its advantages over traditional convolution operations. Here is the revised content for Section 2.3.2:

      In traditional convolution operations, the shape and size of the convolutional kernels are fixed, which limits their flexibility in handling different image features. To overcome this limitation, this study introduces an adaptive convolution structure (AKConv) that can dynamically adjust the shape of the convolutional kernel based on the features of the input image. This dynamic adjustment not only enhances the model's adaptability to different image features but also reduces the number of model parameters and computational load.

      The core advantage of AKConv lies in the dynamic adaptability of its convolutional kernels. Specifically, AKConv first performs convolution operations using an initial sampling shape preset based on the features of the input image. Subsequently, the shape of the convolutional kernel is dynamically adjusted according to the offset learned from the image features to adapt to changes in image characteristics. This dynamic adjustment mechanism allows AKConv to flexibly adapt to targets of different sizes and shapes, thereby achieving higher precision in the feature extraction process.

      Furthermore, the design of AKConv allows it to automatically adjust the sampling shape of the convolutional kernel based on image features, optimizing the number of parameters and computational efficiency. This is particularly important in the construction of lightweight models, as it helps reduce the complexity of the model while maintaining or improving performance. AKConv, through its innovative kernel design, achieves a linear growth in the number of parameters, significantly reducing the computational burden of the model, making it suitable for various hardware environments, including mobile devices and embedded systems.

 

(5) Paragraph 3.4.2: Authors must include more details about how the external validation was conducted. Were there significant differences between the external validation dataset and the original dataset in terms of pest appearance, lighting, or other environmental conditions?

Modification instructions: Thank you for your review comments on our paper. Following your suggestions, we have added detailed information about external validation in Section 3.6 of the paper. Below are the modifications we made to Section 3.6, as well as a specific explanation of the differences between the external validation dataset and the original dataset.

      In Section 3.6.1, we have described in detail the experimental setup for external validation. We ensured that the external validation dataset significantly differs from the original dataset in the following aspects to test the model's generalization ability:

1. Pest appearance: The pest samples in the external validation dataset include more individuals with defects and at different growth stages, which contrasts with the relatively uniform and intact pest samples in the original dataset.
2. Lighting conditions: The images in the external validation dataset were captured under various lighting conditions, including overcast, rainy days, and low-light environments during dawn and dusk, while the original dataset was primarily captured under normal lighting conditions.
3. Pest quantity: The number of pests in the images of the external validation dataset varies, which contrasts with the more uniform pest distribution in the original dataset.

      In Section 3.6.2, we present the detailed results of the external validation and compare the performance of different models under these challenging conditions. We found that despite the differences in pest appearance, lighting, and background between the external validation dataset and the original dataset, the YOLOv8-SAB model still demonstrated high accuracy and robustness. These results further prove the effectiveness and practicality of the model. Additionally, we have provided a detailed table (Table 4) showing the performance metrics of different models in external validation, including Precision, Recall, mAP values, and F1 scores. These data further support our conclusion that the YOLOv8-SAB model significantly outperforms other models in pest detection. We believe that these modifications and additions have fully addressed your review comments and strengthened the persuasiveness of the paper. We look forward to your further feedback.

 

(6) General comment: Improve figure captions and explanations for mathematical equations.

Modification instructions: In response to your suggestions regarding the improvement of chart titles and explanations of mathematical equations, we have carefully revised the paper. Here are the specific modifications we have made:

1. Improvement of chart titles: We have ensured that all chart titles are descriptive enough to clearly convey the core content of the chart independently of the text. Each chart title is written in concise and clear language, using descriptive terms to emphasize the results of the data in the chart or the main findings.
2. Explanation of mathematical equations: For the mathematical equations in the paper, we have added more detailed explanations and step-by-step instructions. Now, each equation has a clear explanation that describes the meaning of the variables, the source of the equation, and their role in the research, so that readers can more intuitively understand the relationships expressed by the equations.
3. Correspondence of charts and equations: We have checked all charts and equations in the text to ensure they are closely related to the content discussed and are clearly referenced in the text.

 

(7) All other comments you can find in pdf document.

     The points that need to be revised in the document are as follows:

1. YOLO (You Only Look Once are widely employed in the field of crop disease and pest detection)

     Modification instructions: Have been changed in accordance with the advice given.Changed to YOLO (You Only Look Once)

 

2. How important is the speed of getting results here, if accuracy is more important then some "slower" algorithms might give better results.

     Modification instructions: Thank you for reviewing our paper. In the model improvement, we adopted the YOLOv8 network, which demonstrates exceptional performance in detection speed and accuracy. This network structure is specifically designed for efficient and accurate object detection, capable of achieving high accuracy while maintaining rapid detection speeds, meeting the needs of real-time processing. Through carefully optimized algorithms and adjustments to the network structure, YOLOv8 ensures efficiency and accuracy in a variety of practical application scenarios.

 

3. Why those? where in the text we can found information about detection accuracy compared byphones (cameras). also where we can found technical information (specification) about camera

     Modification instructions: Thank you for your valuable feedback. Therefore, we have revised it to: This study uses devices such as the iPhone 14 Pro Max (Apple Inc., Cupertino, United States) and the Redmi K50 (Xiaomi Inc., Beijing, China) for data collection, in order to enhance the robustness and generalization capability of the recognition model, enabling it to adapt to various shooting conditions and devices.

 

4. What selections? how were the images chosen?

     Modification instructions: Have been changed in accordance with the advice given.The revised content is as follows: After applying image enhancement techniques to the initial dataset images, a dataset containing 6,442 images was formed. Subsequently, 6,198 images were selected as the final dataset based on the presence or absence of pest targets and without distortion, to ensure the quality and diversity of the data.

5. Text in figure 3 is not readable. Also it would be nice to have an example of at least one representative image with labeling box(s), also explain why all those visual features is interesting for me or other readers

     Modification instructions: In response to the issue of readability concerning Figure 3 that you raised, we have taken the following measures to optimize it:

1. We have transferred the legend text of Figure 3 to the main text section for a detailed explanation, allowing readers to more clearly understand the results of the image target label information.
2. We have reviewed and adjusted the label text in all images to ensure that the font is clear and easy to read. Particularly for Figure 3, we have increased the font size of the labels and improved the layout to enhance overall readability.

 

     6.“SloU optimization loss function” explain more

     Modification instructions: Have been changed in accordance with the advice given.The revised content is as follows: The original loss function has been replaced with the SIoU loss function, which helps the model better learn how to accurately localize pests. The calculation takes into account the center distance of the bounding boxes, as well as the area ratio of the predicted box to the true box, thereby providing a more comprehensive assessment of the quality of the predicted boxes.

 

     7.“YOLOv8 is in tea pest identification and detection, the original YOLOv8 model is affected by factors such as target overlap, occlusion, and missing body parts, resulting in poor recognition performance.” lmprove this sentence.

     Modification instructions: Have been changed in accordance with the advice given.The revised content is as follows: In the context of tea pest identification and detection, the original YOLOv8 model may encounter challenges due to factors such as target overlap, occlusion, and missing body parts, which can lead to suboptimal recognition performance.

 

     8.“Figure 6. Diagram illustrating the structure of the SIoU loss function.” explain what is green and red box

     Modification instructions: Have been changed in accordance with the advice given.The revised content is as follows: Based on this, a prediction is made on one of the X-axis or Y-axis and continuously approached, causing the predicted box to continually converge towards the real box. The structure of its SIoU loss function is shown in Figure 5. The green box in the figure represents the real box, and the red box represents the predicted box.

 

     9.“Adaptive initial sampling shapes.”explain more

     Modification instructions: Have been changed in accordance with the advice given.The revised content is as follows: Additionally, the design of AKConv allows it to automatically adjust the sampling shape of the convolutional kernel based on the density of pests and the phenotypic characteristics of the targets, thereby optimizing the number of parameters and computational efficiency. The adaptive sampling shape of AKConv is first determined by its coordinate generation algorithm, which establishes the initial sampling positions of the kernel. These positions can dynamically change according to the features and targets in the image. Subsequently, to better adapt to the size and shape variations of the targets in the image, AKConv adjusts the sampling positions of the kernel based on the characteristics of the targets. Finally, the feature map is resampled according to the adjusted sampling shape to achieve more precise feature extraction.

 

10. if you have all three, then it can be added explanation about TN

    Modification instructions: Thank you for your review comments on our paper. In the revised manuscript, we have removed the original Figure 12 (Figure 12. Binary classification confusion matrix.), and therefore we believe it would be appropriate to introduce TP, FP, and FN in the formula. If you think it would be better to include TN as well, we will make the amendment in the next revision. Thank you for your meticulous and conscientious review.

 

11. Comment related to this figure and figure 2 andfigure 6 and others, is question for those who dont know what is tea plantation and how it looks likeWhat is this yellow? it looks like lemon... also infigure 16 there is something green, so please explain

    Modification instructions: Thank you for your review comments. We recognize that traditional pest management methods are more effective in identifying larger pests, but they fall short in capturing and recording detailed characteristics of minute pests. To address this issue, our study employed a macro lens to directly photograph the tiny pests on tea leaves and used yellow sticky traps to attract the pests. Once a sufficient number of pests had gathered on these yellow sticky traps, we used the macro lens to collect detailed images. The yellow parts in the figures represent the sticky traps, while the green parts denote the tea leaves. The application of these techniques significantly enhanced our ability to capture and identify minute pests and facilitated a more precise analysis of their characteristics. The revised manuscript now includes these detailed explanations.

 

12. typing error light linght

    Modification instructions: Have been changed in accordance with the advice given.

Reviewer 4 Report

Comments and Suggestions for Authors

This paper proposes an approach based on YOLOv8 for tea pest detection. Here are some considerations.

 

1. The introduction may be improved by adding references to similar works which exploited YOLOv8 in related fields. For example, check https://doi.org/10.1016/j.compag.2024.108728 or https://doi.org/10.1016/j.aiia.2024.02.001.

2. The authors should specify the overall number of labels.

3. Figure 4 is taken from https://scikit-learn.org/stable/_images/grid_search_cross_validation.png.

4. The authors should specify the model density used in Table 2 for complete reproducibility. Furthermore, the authors should highlight the best results achieved by their proposal in the table, for example, by using bold characters. The authors should also provide values for mAP 0.5 and mAP 0.95. Finally, the values for layers and gradients provide low value and can be removed. Model size and inference speed are much more interesting, especially for deployability on constrained devices.

5. The interpretability analysis performed via Grad-CAM should be extended. To provide a complete assessment, the authors should also provide negative samples, i.e., samples where the proposed approach underperforms.

6. The authors should also evaluate newer YOLO models, specifically YOLOv9 and YOLOv10. Without this comparison, the effectiveness of the proposal can be questioned.

7. The authors should provide hints on the applicability of their proposal in real, constrained devices. Furthermore, a wider perspective on future works and the limitations of the current work should be provided. Finally, if possible, the dataset should be provided for reproducibility.

For all these reasons, I suggest a major revision before the paper is considered for publication.

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) The introduction may be improved by adding references to similar works which exploited YOLOv8 in related fields. For example, check https://doi.org/10.1016/j.compag.2024.108728or https://doi.org/10.1016/j.aiia.2024.02.001.

Modification instructions: Thank you wholeheartedly for your valuable suggestions. Following your guidance, we have improved the introduction section by adding citations of research works utilizing YOLOv8 in the relevant field. This not only strengthens the academic background of our study but also highlights the uniqueness and innovation of our work. We believe that these additional citations will help readers better understand the application prospects of YOLOv8 in the field of agricultural phenotyping, as well as how our research innovates and improves upon the existing foundation. Thank you again for your suggestions, and we look forward to your further guidance.

 

(2) The authors should specify the overall number of labels.

Modification instructions: Thank you for your valuable feedback. We have added a total of 7104 labels to the revised manuscript as suggested.

 

(3) Figure 4 is taken from https://scikit-learn.org/stable/_images/grid_search_cross_validation.png.

Modification instructions: Thank you for pointing out the issue in our paper. We realized there was an oversight in the use of Figure 4, and we have recreated Figure 4 (now Figure 3 in the revised manuscript) to ensure that all images are generated by us and are in complete accordance with our research content. We have conducted a thorough check of all images and tables in the paper to ensure that everything meets academic standards.

 

(4) The authors should specify the model density used in Table 2 for complete reproducibility. Furthermore, the authors should highlight the best results achieved by their proposal in the table, for example, by using bold characters. The authors should also provide values for mAP 0.5 and mAP 0.95. Finally, the values for layers and gradients provide low value and can be removed. Model size and inference speed are much more interesting, especially for deployability on constrained devices.

Modification instructions: Thank you sincerely for your professional comments on our paper. Your suggestions are of great guiding significance to our research work. Based on your advice, we have carefully revised and supplemented the paper. Below are the responses to the specific issues you raised:

1. Regarding model density, the specific configurations and parameters used in Table 2 of the revised manuscript are detailed to ensure that other researchers can fully reproduce the experimental results. Detailed information including the model structure selection, number of parameters, and hierarchical structure has been included.
2. To highlight the best results of our method, we have used bold font in Table 2 to display the best performance in each category, namely the data related to the YOLOv8-SAB model.
3. We have also added the values of mAP50, mAP 0.95, and FPS in Table 2 to more comprehensively evaluate the model performance.
4. Following your suggestion, we have removed the number of layers and gradient values from the table, as these values are relatively low, and you pointed out that model size and inference speed are more important for deployment on constrained devices. We retained the values for the number of parameters and GFLOPs because these metrics are crucial for assessing the practicality and deployability of the model. Regarding model density,
5. Additionally, we have added a discussion on the structural advantages of the YOLOv8-SAB model in the methods section of the paper, especially its effectiveness in handling complex scenes and its performance in object detection tasks from the perspective of smart agricultural devices.

 

(5) The interpretability analysis performed via Grad-CAM should be extended. To provide a complete assessment, the authors should also provide negative samples, i.e., samples where the proposed approach underperforms.

Modification instructions: We sincerely appreciate your review of our paper and the valuable feedback provided. Your suggestions are crucial for improving the content of our paper. In response to your request for an expanded Grad-CAM explainability analysis, we have made the following revisions:

      We have added a Grad-CAM analysis of the YOLOv8-SAB model's performance on negative samples. By doing so, we not only demonstrate the model's effectiveness on correctly classified samples but also reveal potential limitations when the model is influenced by certain non-target features.

      In the revised paper, we provide a detailed description of the Grad-CAM analysis results of the YOLOv8-SAB model on negative samples. We found that although the model exhibits some bias when recognizing certain non-target features, its accuracy in localizing the actual pest areas remains high.

      We further emphasize the performance of the YOLOv8-SAB model on negative samples and compare it with other models in the ablation study to demonstrate the robustness and superiority of our model.

 

(6) The authors should also evaluate newer YOLO models, specifically YOLOv9 and YOLOv10. Without this comparison, the effectiveness of the proposal can be questioned.

Modification instructions: Thank you for your valuable comments on our research. Following your suggestions, we have evaluated the YOLOv10 and YOLOv9 models and compared them with our proposed YOLOv8-SAB model. Here are our revised content and responses:

1. Model comparison experiments: We have added a comparison experiment of YOLOv10, YOLOv9, and YOLOv8-SAB models in Section 3.6. The experimental results show that YOLOv8-SAB outperforms YOLOv10, YOLOv9, and the original YOLOv8 model in terms of AP values for four pests and overall mAP, as shown in Table 3. This demonstrates the superior performance of YOLOv8-SAB in pest detection.
2. Model detection experiments: In Section 3.6.1, we tested the detection performance of YOLOv8-SAB, YOLOv10, and YOLOv9 models under different lighting conditions and levels of pest body integrity. The results indicate that YOLOv8-SAB performs the best under these critical conditions, with higher confidence and accuracy.
3. External validation comparison: In Section 3.6.2, we assessed the performance of the YOLOv8-SAB model on an external validation dataset. Compared to YOLOv10, YOLOv9, and other models, YOLOv8-SAB shows significant improvements in Precision, Recall, mAP, and F1 metrics, as shown in Table 4. This further proves the model's generalization capability and practicality.

 

(7) The authors should provide hints on the applicability of their proposal in real, constrained devices. Furthermore, a wider perspective on future works and the limitations of the current work should be provided. Finally, if possible, the dataset should be provided for reproducibility.

Modification instructions: Thank you for your valuable feedback on this research. We have carefully considered your suggestions and have made corresponding revisions and additions to the conclusion section of the paper. Below are the responses to the specific issues you raised:

1. Regarding the feasibility of application on actual constrained devices: We have added an analysis of the model's feasibility for application on actual constrained devices in the discussion section. The improved YOLOv8 network model we proposed has taken into account computational efficiency and model size in its design, making it suitable for operation on resource-constrained devices. For instance, by incorporating AKConv and BiFormer, we have reduced the number of model parameters while maintaining high accuracy, thereby decreasing the demand for computational resources. Furthermore, we have explored how to further optimize the model structure and algorithms to meet the needs of edge computing and Internet of Things devices.
2. Regarding future research directions and the limitations of the current work: We have clearly outlined the limitations of this study and proposed potential directions for future research in the discussion section. Although our model has shown excellent performance in tea pest detection, there are still limitations in terms of environmental adaptability, pest species diversity, and the size of the dataset. Future research could explore model testing under cross-regional and multi-environmental conditions, increase the variety of pest samples, and investigate more efficient model optimization strategies to enhance the model's generalization and adaptability.
3. Regarding the provision of the dataset: To ensure the reproducibility of the research, we have organized the dataset used for training and testing the model and made it available upon request from the corresponding author. The dataset contains images of various types of tea pests, along with the corresponding annotation information, to support other researchers in verifying and comparing the effectiveness of different methods.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors, much of the text is in red indicating that it has been altered. However, the changes were not made in a manner pertinent to what was requested. For example, the captions are still poor and do not reflect all the elements that appear in the figures and tables. The materials and methods section still contains little description of how it was carried out and the correct sample size. The discussion section has not yet been improved to discuss all the data raised and presented in the results section. I counted only 3 bibliographic references in the entire section. The authors should write an adequate and pertinent discussion both in relation to the manuscript and for the journal’s readers. In this sense, my opinion is to reject the manuscript and that the authors rewrite it in an adequate and pertinent manner, but a new possibility can be Major Revision.

Comments on the Quality of English Language

Please, check all manuscript in grammar, spelling and verbosity.

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) Dear Authors, much of the text is in red indicating that it has been altered. However, the changes were not made in a manner pertinent to what was requested. For example, the captions are still poor and do not reflect all the elements that appear in the figures and tables. The materials and methods section still contains little description of how it was carried out and the correct sample size. The discussion section has not yet been improved to discuss all the data raised and presented in the results section. I counted only 3 bibliographic references in the entire section. The authors should write an adequate and pertinent discussion both in relation to the manuscript and for the journal’s readers. In this sense, my opinion is to reject the manuscript and that the authors rewrite it in an adequate and pertinent manner, but a new possibility can be Major Revision.

Modification instructions: Thank Thank you for your review of the manuscript and your valuable comments. We have carefully considered your feedback and made the following revisions in response to the issues you raised:

       1. Figure caption issues: We have thoroughly revised the captions for all figures to ensure they accurately reflect all elements in the figures and provide more complete and clear information.

       2.Materials and Methods section: We greatly appreciate your suggestions. To provide a clearer view of how we improved the model, this study detailed the structure of our improved YOLOv8 network in Figure 4. In the network architecture, we replaced the standard convolutional layers (Conv) with attention kernel convolutional layers (AKConv), achieving dynamic weighted optimization of the feature extraction process through module replacement at the code level. The integration of the BiFormer module and inter-layer connections embed the BiF module in the pre-sequence layer before the detection head in the forward path of the object detection process. By precisely configuring the network structure, we ensure the output of this module is smoothly passed to the detection head and receives input feature maps from previous layers to optimize multi-scale feature fusion and efficient representation. For the improvement of the loss function, modifications were made directly in files such as loss.py. If necessary, we will add more details later. Regarding the sample size of the image dataset, this study collected a total of 1,346 original images with a resolution of 7952x5340 in .JPG format. From the original images, we selected 189 high-quality Xyleborus fornicatus Eichhoff images, 221 Empoasca pirisuga Matumura images, 168 Arboridia apicalis images, and 225 Toxoptera aurantii images to form the initial dataset. To enhance the model's learning and generalization capabilities and to strengthen the model's robustness, the initial dataset images were processed using image enhancement techniques to form a dataset containing 6,442 images. Subsequently, images without pest targets and those with more than 80% missing targets were removed, resulting in a final dataset of 6,198 images to ensure data quality and diversity. We have added detailed descriptions of the above sections to improve the transparency and reproducibility of this part of the content.

      3.Discussion section: We have completely rewritten the discussion section to ensure it comprehensively analyzes and discusses all data presented in the results section. We have added in-depth analysis of the results and compared them with existing literature to demonstrate how our research contributes to the field.

      Here is the revised 'Discussion' section:

4. Discussion

      Pest infestations in tea gardens are one of the common issues in the process of tea cultivation. To achieve rapid and accurate identification of early-stage minor pests in tea gardens, this study proposes an improved YOLOv8 network model for tea pest detection, addressing issues such as small datasets and difficulty in extracting target pest phenotypic features in the tea pest detection process. The model significantly enhances detection accuracy and robustness by introducing advanced technologies such as the SIoU loss function, AKConv, and BiFormer. Experimental results show that the improved model has achieved a precision rate of 98.16% in tea pest detection tasks, which is a 2.62% increase compared to the original YOLOv8 network, demonstrating its efficiency and accuracy in tea pest detection. This improvement significantly enhances the model's ability to recognize minor pests, which is of great significance for ensuring the healthy development of the tea industry, especially in the context of Yunnan's ecological tea industry.

       The improved YOLOv8 network model in this study not only enhances the model's learning ability for small target pest samples in tea gardens but also improves the acquisition capability of target location information, thereby enhancing the model's perceptual performance for targets. Compared with existing studies, it is noted that Solimani et al. [8] proposed a lightweight YOLOv8n-ShuffleNetv2-Ghost-SE model, which achieved an average precision of 91.4% and a recall rate of 82.6%, while this study has improved the average precision and recall rate by 6.76% and 15.45% respectively. Fuentes et al. [18] proposed a tomato pests and disease detection network that combines VGG and ResNet as deep feature extractors. However, the network model detected an average precision of only 85.98%, which is 12.18% lower than that of this study, indicating a higher detection accuracy in this research. Dai et al. [19] introduced the Swin Transformer mechanism and improved feature fusion strategy into the YOLOv5m model, achieving a recall rate of 93.1%, an F1 score of 94.38%, and an average precision of 96.4%, but this study has significantly improved by 4.95%, 3.07%, and 1.76% respectively. He[30] and others proposed a tea garden pest recognition method based on an improved YOLOv7 network, using the MPDIoU optimized loss function: which improved the model's convergence speed and simplified the calculation process; applied spatial and channel reconstruction convolution to reduce feature redundancy, reducing model complexity and computational cost; introduced a dual-route attention visual transformer, enhancing the model's computational distribution flexibility and content-aware capability. The improved YOLOv7 model saw an increase in Precision, Recall, F1, and mAP compared to the original YOLOv7 by 5.68%, 5.14%, 5.41%, and 2.58%, respectively. However, the improved YOLOv8 model in this study effectively increased accuracy while reducing the number of parameters. Compared to the original YOLOv8 model, the improved YOLOv8 model detected four types of tea pests with AP values of 98.16%, 98.32%, 98.06%, and 98.03%, respectively, increasing by 2.53%, 2.76%, 2.69%, and 2.43%, with an overall mAP increase of 2.62%. After external data validation, the improved YOLOv8 network model's Precision was increased by 8.12%, 7.40%, 4.52%, and 15.98% compared to YOLOv10, YOLOv9, YOLOv8, and YOLOv7, respectively. Recall was increased by 13.54%, 6.59%, 3.01%, and 9.68%, respectively, mAP values were increased by 3.04%, 4.27%, 2.98%, and 5.56%, respectively, and the balance score F1 was increased by 10.87%, 7.00%, 3.78%, and 13%, respectively. Yang[45] and others proposed a tea garden pest detection model based on an improved YOLOv7-Tiny algorithm. After adding Biformer to the original YOLOv7 model, the model's mAP0.5 increased from 88.6% to 91.6%. In this study, after adding Biformer to the original YOLOv8 model, the model's mAP0.5, Precision, and Recall were 96.72%, 94.59%, and 95.54%, respectively, significantly higher than the YOLOv7-Tiny model for tea garden pest detection. The YOLOv7-Tiny model only had an average accuracy of 93.23%, which is 4.93% lower than the accuracy of pests detected in this study. The improved YOLOv8 network model in this study has shown stronger performance in feature extraction and target localization for tea pest detection, achieving higher detection accuracy through specially optimized loss functions and network structures. In addition, by introducing AKConv and BiFormer, the model has shown stronger performance in feature extraction and target localization, which are less involved in existing studies.

       Although this study has performed well in the identification and detection of specific foreign targets such as minor tea pests, there are still certain limitations. First, the model's performance has mainly been tested under specific environmental conditions in Yunnan tea gardens, and its performance in other regions and different environmental conditions has not been fully verified, so there is a limitation in environmental adaptability. Secondly, the main types of pests studied are relatively limited, and the model's recognition ability and accuracy for a wider range of pest types still need further testing and optimization. In addition, due to the limited scale of the training dataset, this may limit the model's generalization ability in recognizing new types of pests. To enhance the model's generalization, it is recommended to conduct more in-depth data augmentation processing on the dataset in this study. Specifically, the diversity of the dataset can be enriched by using techniques such as generative adversarial networks, feature fusion, oversampling, and undersampling, thereby enhancing the model's recognition ability for different pest types. Based on this, the model's performance may be affected by environmental changes, the diversity of pest types, and dataset biases. Therefore, potential future directions can be developed from the above content to further promote the development of tea pest detection technology and provide scientific and technological support for the sustainable development of the tea industry.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors successfully fixed all the highlighted issues. Therefore, the paper can be considered for publication.

Author Response

Thanks very much for your time to review this manuscript. I really appreciateyou’re your comments and suggestions. We have considered these comments carefully and triedour best to address every one of them.

(1) The authors successfully fixed all the highlighted issues. Therefore, the paper can be considered for publication.

Modification instructions: We sincerely thank you for your review and positive assessment of our manuscript. We are pleased to learn that you believe we have successfully resolved the previously identified issues, and that the paper is now ready for consideration for publication. We are committed to ongoing efforts to ensure that our research adheres to the highest academic standards and contributes meaningfully to the scholarly community. Thank you again for your valuable time and expert opinion.