Research on the Shape Classification Method of Rural Homesteads Based on Parcel Scale—Taking Yangdun Village as an Example

Round 1

Reviewer 1 Report

This paper analyzes the form classification of Yangdun Village homestead. This is a very meaningful study. But there are some unclear places, I hope the author can be clear.

1. This research topic is quite general. What is the innovation? How can it be implemented in different situations?

2. The summary looks good, but more than 50% of it is about the results.

3. In the introduction section, documents should be logically organized, rather than listed one by one.

3. The part of the research method is not very clear. In a sentence or two, tell the reader how important the method is for the research and why it would be helpful to use it. How do you know it won't give you better results any other way?

4. The analysis in the conclusion is shallow. Further clarification is needed. The contribution and significance of this study also need to be discussed.

5. Check the order and format of your references.

 Minor editing of English language required

Author Response

We sincerely appreciate your comprehensive and insightful comments and suggestions. Your feedback has been invaluable in enhancing the quality of our research. Here are our responses to your specific points:

1. This research topic is quite general. What is the innovation? How can it be implemented in different situations?

Reply: Thank you for raising this point. We have clarified the innovation aspects of our research in the introduction section. We highlight the following innovations:

(1) Innovation in Research Content. While studies on shape recognition of urban buildings and rural settlements exist, our focus on shape classification of rural homesteads is relatively unexplored.

(2) Innovation of Research Methods: We distinguish our research by comparing multiple feature extraction methods within a single classification algorithm, as opposed to previous studies that often compared various classification methods with a fixed feature extraction method. Furthermore, we undertake a comparison between machine learning and deep learning approaches, shedding light on the application effectiveness of common shape classification methods in homesteads through quantitative analysis. We emphasize that our research framework can be adapted to different regions or countries, with necessary adjustments made to local parameters based on available data.

2. The summary looks good, but more than 50% of it is about the results.

Reply: We appreciate your feedback on the summary section. Following your guidance, we have significantly reduced the emphasis on the results in the summary section. The updated conclusions now reflect the key findings of our research, focusing on the most important outcomes and their implications for rural land management. The updated conclusions are as follows:

The conclusions are as follows: First, we found that the method combining Fast Fourier Transform (FFT) and Random Forest (RF) proved to be more suitable for homesteads shape classification, achieving an average accuracy rate of 88.6%. Second, the combination of multiple feature extraction methods did not lead to improved recognition accuracy. Specifically, the accuracy of the FFT+HIM combination was 88.4%, while the accuracy of FFT+HIM+BCSI was 88%. Third, it should be noted that the BCSI, although commonly used in urban contour classification, proved unsuitable for homesteads, yielding an average accuracy rate of only 58%. There was no precise numerical correlation between the category and shape index in homesteads. Fourth, it is worth noting that over half of the homesteads in Yangdun Village had a rectangular shape. In the five years preceding and following the "homesteads reform", there was a significant increase in the number of regular-shaped homesteads being vacated, while irregular-shaped homesteads exhibited an upward trend. The method we have proposed has the potential to assist villages in China and globally in swiftly assessing the level of homesteads regulation.

3. In the introduction section, documents should be logically organized, rather than listed one by one.

Reply: Thanks for the suggestion. We have restructured the introduction section to ensure a more logical flow. The first category of methods, based on contour analysis, is now presented in a separate paragraph. We provide a concise overview of the four methods it encompasses, followed by examples of each method, along with their respective strengths and limitations. Additionally, we introduce the two applications of machine learning, highlighting their drawbacks and leading into the discussion of deep learning.

4. The part of the research method is not very clear. In a sentence or two, tell the reader how important the method is for the research and why it would be helpful to use it. How do you know it won't give you better results any other way?

Reply: Thanks for reminding us. We have addressed your concern by adding an explanatory section at the beginning of the Materials and Methods section. This section now outlines why we chose the specific methods employed in our study, emphasizing their relevance in shape classification of homesteads. We acknowledge that it is impossible to definitively determine the performance of non-comparative methods, which we have duly mentioned in the limitations section, paving the way for future research. The updated content in the article is as follows:

This article employs various methods: Fast Fourier Transform (FFT), Hu Invariant Moments (HIM), Boyce and Clark Shape Index (BCSI), and AlexNet. These methods, respectively representing image contour, image region, urban boundary, and deep learning approaches, have been effectively applied in extracting shapes of urban boundaries and rural buildings. However, determining the most suitable single or combined method for classifying homesteads shapes required a comprehensive investigation in this study.

5. The analysis in the conclusion is shallow. Further clarification is needed. The contribution and significance of this study also need to be discussed.

Reply: Thanks for the suggestion. We have revisited the conclusion section and condensed the description of results while providing more comprehensive analysis and discussion in the relevant section. In addition, we have included a discussion on the contribution and significance of our study, emphasizing its relevance to rural land management, planning, and optimization. The significance of the updated article is as follows:

Building on the preceding discussion, numerous studies have focused on methods for building extraction in urban and rural settings, yet relatively few have delved into the classification of homesteads vector graphics. The comparative analysis of various feature extraction methods presented in this article offers valuable insights for related research. Furthermore, this study incorporated established methods from other domains into the realm of homesteads classification. Our proposed method effectively categorizes homesteads shapes, reducing the need for time-consuming manual evaluations and mitigating the variability introduced by cognitive differences. This approach facilitates managers in promptly understanding and adjusting homesteads regularities, with broad applications in rural homesteads investigation and management. Moreover, it holds significance as a reference for international rural planning.

6. Check the order and format of your references.

Reply: We have meticulously reviewed and revised the order and format of our references to ensure compliance with the appropriate style guidelines. Some frequently cited articles are now presented with their first serial number to maintain consistency.

Once again, we express our gratitude for your constructive feedback, which has significantly improved the overall quality of our research. We look forward to any further insights you may provide during the review process.

Author Response File: Author Response.docx

Reviewer 2 Report

This article simply organizes and compares existing models for extracting and classifying the shapes of rural homesteads. The methodology lacks novelty and application value.  The problems are as follows:

1. In line 112, the author mentioned that rural homesteads are more complex and changeable in shape compared with urban buildings. I don't agree with this point of view. Urban architectures are more various, modern and creative comparing to rural homesteads. The researches of extraction, layout and organization of urban buildings have been rich. Thus, I’m not sure the significance of the extraction and classification of rural homestead shape. 

2. In line 113, the author also said ”no research on the shape of rural homesteads has been found”, which is very confusing. The following literatures about rural homesteads is found. A supplement of relevant literatures is needed.

[1]Meng, C., Song, Y., Ji, J. et al. Automatic classification of rural building characteristics using deep learning methods on oblique photography.  Building Simulation. 2022, 15, 1161–1174.

[2]Wei R, Fan B, Wang Y, et al. MBNet: Multi-Branch Network for Extraction of Rural Homesteads Based on Aerial Images[J]. Remote Sensing, 2022, 14(10): 2443.

[3]Zhou, J.; Liu, Y.; Nie, G.; Cheng, H.; Yang, X.; Chen, X.; Gross, L. Building Extraction and Floor Area Estimation at the Village Level in Rural China Via a Comprehensive Method Integrating UAV Photogrammetry and the Novel EDSANet. Remote Sens. 2022, 14, 5175. 

3. According to local policy, demolition is one of the main driver for the homesteads reform. According to the changes of homesteads in study area (Figure 11), the homesteads in the northern part of the village collectively disappeared, while the remaining homesteads didn’t present changes obviously. Therefore, what is the significance of exploring the temporal  changes in the shape of the homesteads? 

4. In line 152, Google Earth satellite images are used for extracting the shapes of homesteads. However, it is known that buildings will be a little distorted due to shooting angles of images and height difference of buildings. The distortion will affect the extraction and classification of the homesteads’ shapes. Did the author considered the effect of distortion? 

5. In figure 1, triangle-like or trapezoid-like homesteads exist along a diagonal road in area 2. Please explain why there are no triangle-like or trapezoid-like types in the shape categories. 

6. In line 169, the author said “ Used fast Fourier transform(FFT), HIM and BCSI to extract the shapes of each homestead, and divided them into single feature and multiple combination features”. Please give an example of what is a multiple combination feature and how a homestead is divided into a multiple combination feature.  

7. In line 201, the evidence of the setting of FFT descriptor should be more persuasive. In line 246, the explanation of the selection of optimal feature point is unclear, and the descriptions of the sub-figures in Figure 6 should be supplemented. 

8. The sample size for RF is a bit small. In line 271, the explanations of OOB and numbers of decision trees(ntrees) are inconvincible. 

9. Please explain the production of the true label in the part Between-category Accuracy.  

English should be improved. 

Author Response

There is a format disorder here. We recommend reading the attachment.

We would like to express our sincere gratitude for your thorough and insightful comments and suggestions. Your feedback has been instrumental in improving the quality of our research. Here are our responses to your specific points:

This article simply organizes and compares existing models for extracting and classifying the shapes of rural homesteads. The methodology lacks novelty and application value. The problems are as follows:

1. In line 112, the author mentioned that rural homesteads are more complex and changeable in shape compared with urban buildings. I don't agree with this point of view. Urban architectures are more various, modern and creative comparing to rural homesteads. The researches of extraction, layout and organization of urban buildings have been rich. Thus, I’m not sure the significance of the extraction and classification of rural homestead shape.

Reply: We appreciate your input regarding the comparison of rural homestead shapes with urban buildings. We have modified the text to provide a more accurate representation of the differences. Specifically, we now emphasize that rural homesteads are more diverse in composition, scattered, and exhibit greater variability in shape after completion compared to urban buildings. We have also expanded upon the significance of studying rural homestead shapes, particularly in the context of land use, planning, and optimization.

Finally, the research significance has been revised as follows:

Building on the preceding discussion, numerous studies have focused on methods for building extraction in urban and rural settings, yet relatively few have delved into the classification of homesteads vector graphics. The comparative analysis of various feature extraction methods presented in this article offers valuable insights for related research. Furthermore, this study incorporated established methods from other domains into the realm of homesteads classification. Our proposed method effectively categorizes homesteads shapes, reducing the need for time-consuming manual evaluations and mitigating the variability introduced by cognitive differences. This approach facilitates managers in promptly understanding and adjusting homesteads regularities, with broad applications in rural homesteads investigation and management. Moreover, it holds significance as a reference for international rural planning.

2. In line 113, the author also said "no research on the shape of rural homesteads has been found", which is very confusing. The following literatures about rural homesteads is found. A supplement of relevant literatures is needed.

[1]Meng, C., Song, Y., Ji, J. et al. Automatic classification of rural building characteristics using deep learning methods on oblique photography.  Building Simulation. 2022, 15, 1161–1174.

[2]Wei R, Fan B, Wang Y, et al. MBNet: Multi-Branch Network for Extraction of Rural Homesteads Based on Aerial Images[J]. Remote Sensing, 2022, 14(10): 2443.

[3]Zhou, J.; Liu, Y.; Nie, G.; Cheng, H.; Yang, X.; Chen, X.; Gross, L. Building Extraction and Floor Area Estimation at the Village Level in Rural China Via a Comprehensive Method Integrating UAV Photogrammetry and the Novel EDSANet. Remote Sens. 2022, 14, 5175.

Reply: Thank you for your diligent literature search and recommendations. Articles 1 and 3 that you mentioned focus on extracting rural buildings from remote sensing images, rather than addressing the diverse composition of homesteads. It's worth noting that rural buildings are indeed part of rural homesteads, which also encompass courtyards, grain-basking fields, toilets, and cowsheds—elements that supplement the definition of homesteads. Article 2, published by our team, specifically targets the extraction of vector data related to homesteads from remote sensing images. In our study, we delve deeper into the classification of two-dimensional homestead images, comparing four methods and eight scenarios to identify the most suitable classification approach. As such, there is no issue with stating, "No research on the shape of rural homesteads has been found". To prevent any potential misunderstanding, we have elaborated on this point in the introduction, and we have replaced it with the phrase, "there are few studies on the method of classifying the shape of rural homesteads." We appreciate your valuable suggestions, and we have cited the original articles in literature 1 and 2, while also adding the new literature reference 3. For the revised content, please refer to the update below:

In conclusion, numerous studies have examined shape recognition and classification, whether in the context of urban buildings or rural settlements. Traditional feature extraction methods and deep learning approaches are commonly employed. However, rural settlements, in contrast to urban buildings, exhibit greater compositional diversity and dispersion, and their shapes can evolve post-construction. Thus, urban building shape classification methods may not fully encapsulate the complexity of rural homesteads. Rural homesteads encompass more than just buildings; they also include courtyards, grain-basking fields, cowsheds, and other ancillary land. Moreover, not all rural buildings are part of homesteads; for instance, schools, factories, and shops require classification based on land use. Therefore, methods suitable for rural building classification may not necessarily be optimal for rural homesteads classification.

3. According to local policy, demolition is one of the main driver for the homesteads reform. According to the changes of homesteads in study area (Figure 11), the homesteads in the northern part of the village collectively disappeared, while the remaining homesteads didn’t present changes obviously. Therefore, what is the significance of exploring the temporal changes in the shape of the homesteads?

Reply: We appreciate your question regarding the significance of exploring temporal changes in homestead shapes. While it is true that demolition plays a significant role in the reform of homesteads, other changes, such as new construction, expansion, and contraction, also occur naturally over time. The significance of this research lies in understanding the evolving spatial layout of rural homesteads, which has implications for land management, resource allocation, and policy development. By studying these changes, we gain insights into the dynamics of rural land use and can make informed decisions about land planning and optimization.

For example, there were 877 homesteads in Yangdun Village. In 2010, almost all the homesteads in the north of the village were demolished, and a total of 321 homesteads disappeared, with great changes. Although the spatial distribution of other residential sites has not changed greatly, many changes have taken place under the natural evolution, with a total of 72 updates. As shown in the figure, the red color is centralized demolition in the north, and the black color is a non-centralized change in the last 10 years.

4. In line 152, Google Earth satellite images are used for extracting the shapes of homesteads. However, it is known that buildings will be a little distorted due to shooting angles of images and height difference of buildings. The distortion will affect the extraction and classification of the homesteads’ shapes. Did the author considered the effect of distortion?

Reply: Your point about image distortion due to shooting angles and building height differences is valid. In our study, we primarily utilized Google Earth satellite images, which may introduce some level of distortion. However, we found that the actual distortion of rural homesteads, which are typically of low height, is minimal and does not significantly affect our shape extraction and classification methods.

5. In figure 1, triangle-like or trapezoid-like homesteads exist along a diagonal road in area 2. Please explain why there are no triangle-like or trapezoid-like types in the shape categories.

Reply: Thank you for your thorough review. In Area 2, we have indeed identified a homestead with a distinctive regular trapezoidal shape along the diagonal road. This type is relatively uncommon and has been categorized as the third type of irregular rectangle.

6. In line 169, the author said “ Used fast Fourier transform(FFT), HIM and BCSI to extract the shapes of each homestead, and divided them into single feature and multiple combination features”. Please give an example of what is a multiple combination feature and how a homestead is divided into a multiple combination feature.

Reply: We appreciate your request for clarification on multiple combination features. We have included an explanation in the article, distinguishing between single features and multiple combination features. Single features involve using a single algorithm for shape extraction, such as FFT or Hu invariant moments. Multiple combination features, on the other hand, involve combining two or more algorithms for shape extraction, such as combining FFT with Hu invariant moments.

7. In line 201, the evidence of the setting of FFT descriptor should be more persuasive. In line 246, the explanation of the selection of optimal feature point is unclear, and the descriptions of the sub-figures in Figure 6 should be supplemented.

Reply: We have improved the explanations related to the FFT descriptor and feature point selection. We now provide a more detailed example of how multiple feature points are selected and utilized in our analysis. Additionally, we have enhanced the descriptions of the sub-figures in Figure 6 to provide better context and understanding, which is as follows: (a) is the original schematic diagram, where L1, L2, L3 and L4 are rays emitted from the central point；(b) is the effect of taking the internal intersection; (c) is the effect of taking the central intersection; (d) It is the effect of taking external intersection.

8. The sample size for RF is a bit small. In line 271, the explanations of OOB and numbers of decision trees(ntrees) are inconvincible.

Reply: Thanks for your valuable questions. We have taken note of your concerns regarding the sample size for RF and the explanation of OOB. To address this, we have updated the relevant sections, providing more persuasive evidence for the choice of RF parameters and a clearer explanation of the OOB error analysis. The adjusted content was as follows. Thanks again.

When Ntrees iterated to a certain value, the OOB error was insensitive to the change of Ntrees. To avoid the contingency caused by a single test set, 50 random simulation experiments were conducted to mitigate errors (with Ntrees values ranging from 1 to 500). As shown in Figure 7(a), each experiment generated a curve, and it was not easy to find the decision tree threshold from Figure 7(a), resulting in a statistical chart displaying the average values, as illustrated in Figure 7(b). It was observed that once Ntrees reached 120, the OOB error stabilized. Consequently, this value was chosen to optimize the model's processing time, and the default values were selected for other parameters.

Figure 7. The relationship between out-of-bag error and the number of random decision trees

9. Please explain the production of the true label in the part Between-category Accuracy.

Reply: We appreciate your suggestion to explain the production of true labels in the "Between-category Accuracy" section. We have now provided a detailed explanation of how true labels were determined based on the classification standard shown in Figure 3.

Category 1 is Square-like, which means square or square-like, the long side is not more than 1.3 times of the short side, ignoring a small number of bumps or depressions.

Category 2 is rectangular-like, which is rectangular or rectangular-like, consisting of 4 sides, the long side distinctly larger than the short side, with 4 angles, all approximately right-angled, ignoring a small number of bumps or depressions.

Category 3 is irregular rectangular-like, composed of more than 5 sides, and the figure is composed of 2 rectangles.

Category 4 is irregular, there are many bumps in the picture, and the picture is irregular.

Once again, we thank you for your valuable feedback, which has greatly contributed to the refinement of our research. We look forward to any further insights you may provide during the review process.

Author Response File: Author Response.docx

Reviewer 3 Report

Dear Authors,

Thanks for this paper on homesteads shape analysis based on Yangdun village example.

Although you argue that such analysis is important for policies formulation it is not clear for what purposes it is used. The subject of analysis - homested - is not defined. For the understanding of this study the introduction should be updated. So, as the purpose of analysis is unclear, then also applied methods are hardly justified and obtained results hardly interpreted. Neither figures help to understand more, e.g. when prepared carelessly like Figure 1. There is no technical information about input imagery and source. Overall, paper suffers from missing systematic approach to explain problem, describe available data, explain approach to resolve problem, present obtained reults and deliver generalized valid conclusions.

What is meant by a key word "homesteads" here? Is it in line with a common meaning of this word?

Check text - e.g. L120 sentence starts with lower case letter.

Space should precede brackets of references through the text as at L207, L212, not like at e.g. L210, L217, L221, ......)

Text style should be harmonized (e.g. L207).

Author Response

There is a format disorder here. We recommend reading the attachment.

We sincerely appreciate your thorough review and constructive feedback. Your insights have been invaluable in improving the clarity and coherence of our paper. Here are our responses and actions taken regarding your specific points:

1. Although you argue that such analysis is important for policies formulation it is not clear for what purposes it is used. The subject of analysis - homested - is not defined. For the understanding of this study the introduction should be updated. So, as the purpose of analysis is unclear, then also applied methods are hardly justified and obtained results hardly interpreted. Neither figures help to understand more, e.g. when prepared carelessly like Figure 1. There is no technical information about input imagery and source. Overall, paper suffers from missing systematic approach to explain problem, describe available data, explain approach to resolve problem, present obtained reults and deliver generalized valid conclusions.

Reply: We have revised the introduction to offer a more transparent insight into the purpose and significance of our analysis. In addition, we have provided a precise definition of the term "homestead" in our paper, aligning it with its common interpretation, which encompasses the land used by rural villagers for their dwellings and related facilities. Furthermore, we have restructured the paper to adopt a more systematic approach, enhancing the description of the problem, available data, methodology, results, and conclusions. Due to the extensive changes, we kindly invite you to review the introduction, discussion, and conclusion in the revised article for a comprehensive understanding. These sections are not presented here to conserve space.

2. What is meant by a key word "homesteads" here? Is it in line with a common meaning of this word?

Reply: Thanks for your questions. In response to your inquiries, we have incorporated a precise definition of "homestead" in this paper: "Homestead refers to the land utilized by rural villagers for constructing houses and facilities essential to their daily life, encompassing houses, courtyards, grain-basking fields, toilets, cowsheds, and similar components". In China, a country characterized by public ownership, rural homesteads serve as a fundamental welfare provision for farmers and are strictly prohibited from being bought or sold. This definition adheres to the conventional interpretation of the term, signifying the dwelling and adjacent land occupied by a family.

3. Check text - e.g. L120 sentence starts with lower case letter.

Reply: Thanks for the reminder. We have reviewed the entire manuscript to address issues like the lowercase sentence start at L120, ensuring that the text adheres to appropriate grammatical conventions.

4. Space should precede brackets of references through the text as at L207, L212, not like at e.g. L210, L217, L221, ......)

Reply: We have improved the consistency of the references formatting, ensuring that spaces precede brackets throughout the text.

5. Text style should be harmonized (e.g. L207).

Reply: We have harmonized the text style throughout the paper to enhance overall consistency.

We appreciate your careful review and suggestions, which have significantly contributed to the overall quality and readability of our paper. If you have any further comments or questions, please do not hesitate to let us know.

Author Response File: Author Response.docx

Reviewer 4 Report

In this study, the fast Fourier transform (FFT), Hu invariant moments (HIM), BC shape index (BCSI) and AlexNet models were used to classify the shape of homestead plots, and the classification accuracy was compared. The method of the article is reasonable, the results are reliable, and it is a good paper. But there are still some minor issues that need to be improved. (1) The author focuses on China's rural homesteads, so what are the similarities and differences between China's rural homesteads and other countries, and whether the research results of this paper can be used for reference in other countries. Therefore, the author needs to add this kind of description in the introduction and discussion, so as to highlight the research significance of this paper. (2) In the last one or two paragraphs of the introduction, the research objective should be placed at the end of the introduction, and the content needs to be adjusted. (3) Figure 1 suggests adding subtitles and descriptions to the subfigures. (4) Data source, whether it is possible to give the URL of the data source, or tell readers how to obtain the data. (5) Figure 8-10, suggested optimization. (6) Discussions are suggested to be divided into points, and the comparison of methods and sources of errors are added. (7) Significantly reduce the conclusions. If the description is divided into points, two or three lines are reserved for each point.

Minor editing of English language required

Author Response

There is a format disorder here. We recommend reading the attachment.

We are grateful for your thoughtful review and valuable suggestions. Your feedback has greatly contributed to the improvement of our paper. Here are our responses and the actions taken in response to your specific points:

In this study, the fast Fourier transform (FFT), Hu invariant moments (HIM), BC shape index (BCSI) and AlexNet models were used to classify the shape of homestead plots, and the classification accuracy was compared. The method of the article is reasonable, the results are reliable, and it is a good paper. But there are still some minor issues that need to be improved.

1. The author focuses on China's rural homesteads, so what are the similarities and differences between China's rural homesteads and other countries, and whether the research results of this paper can be used for reference in other countries. Therefore, the author needs to add this kind of description in the introduction and discussion, so as to highlight the research significance of this paper.

Reply: We appreciate your suggestion to emphasize the international relevance of our research. We have addressed this by adding a comparison between China's rural homesteads and those in other countries in both the introduction and discussion sections. This comparison highlights the similarities and differences, underlining the broader significance of our findings. The similarities and differences with other countries are as follows:

In both China and Vietnam, rural homesteads are state-owned, with farmers granted the right to utilize them. Given the similarity in homesteads components, farmers are re-quired to adhere to national policies and planning regulations when undertaking home-steads transformations. Rural buildings in Britain are privately owned and can be bought and sold. While these buildings differ from homesteads in China and Vietnam, any alter-ations must align with state regulations or the Localism Act

2. In the last one or two paragraphs of the introduction, the research objective should be placed at the end of the introduction, and the content needs to be adjusted.

Reply: We have restructured the introduction section to place the research objective at the end, as recommended. This helps provide a smoother transition from the background to the purpose of our study.

3. Figure 1 suggests adding subtitles and descriptions to the subfigures.

Reply: In response to your valuable suggestions, we have included subtitles and descriptions for clarity. Here are the details:

1) Descriptions: The homesteads in Yangdun Village exhibit diverse shapes and sizes, as depicted in Figure 1. Notably, homesteads in Area 1 display irregular patterns with noticeable concave and convex features, while homesteads in Areas 2 and 3 exhibit a greater degree of regularity.

2) Subtitles: (a) The geographical location of the study area in China; (b) The specific location of Yangdun Village within Zhejiang Province; (c) The terrain and landform characteristics of Yangdun Village, with yellow rectangular frames delineating distinct areas; (d) Variations in homestead shapes within the study area.

4. Data source, whether it is possible to give the URL of the data source, or tell readers how to obtain the data.

Reply: Thank you for your advice. We have included information about the data source, specifying that the images were obtained from https://earth.google.com and the vector data from the Deqing County Agriculture and Rural Bureau in 2020. This clarification assists readers in understanding the data origin.

5. Figure 8-10, suggested optimization.

Reply: Thanks for the reminder. Borders of the sub-images in Figures 8-10 have been removed to enhance the overall presentation and visual appeal.

6. Discussions are suggested to be divided into points, and the comparison of methods and sources of errors are added.

Reply: Thanks for the suggestion. It has been adjusted as recommended. This makes the discussion section more organized and comprehensive. The updated content includes method comparison (including error sources), research significance and limitations. Please see the discussion of the revised article. The content is long and is not presented here to save space.

7. Significantly reduce the conclusions. If the description is divided into points, two or three lines are reserved for each point.

Reply: Thanks for the reminder. We have significantly reduced the conclusions, following your recommendation. The updated conclusions are now concise, with each point given two or three lines for better clarity. The updated content is as follows:

The conclusions are as follows: First, we found that the method combining Fast Fourier Transform (FFT) and Random Forest (RF) proved to be more suitable for homesteads shape classification, achieving an average accuracy rate of 88.6%. Second, the combination of multiple feature extraction methods did not lead to improved recognition accuracy. Specifically, the accuracy of the FFT+HIM combination was 88.4%, while the accuracy of FFT+HIM+BCSI was 88%. Third, it should be noted that the BCSI, although commonly used in urban contour classification, proved unsuitable for homesteads, yielding an average accuracy rate of only 58%. There was no precise numerical correlation between the category and shape index in homesteads. Fourth, it is worth noting that over half of the homesteads in Yangdun Village had a rectangular shape. In the five years preceding and following the "homesteads reform", there was a significant increase in the number of regular-shaped homesteads being vacated, while irregular-shaped homesteads exhibited an upward trend. The method we have proposed has the potential to assist villages in China and globally in swiftly assessing the level of homesteads regulation.

We sincerely appreciate your feedback and your efforts in helping us improve our paper. If you have any further comments or questions, please do not hesitate to let us know.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Thanks for your detailed answers to these questions. These revisions have effectively emphasized the significance of this study. The most of my questions have been answered. At the same time, the English expression of this article has got an obvious improvement. After this revision, I have the following suggestions:

1. In your revision, a clear definition of a homestead as containing courtyards, grain-basking fields, cowsheds, and other ancillary land has been given out, so that homestead extraction can be distinguished from the conventional extraction of urban and rural buildings. However, it is difficult to find whether the shape extraction includes “special homesteads” which contain these ancillary lands in addition to rural buildings in the study area or the result map . Therefor, please give an example to illustrate this.

2. The figure in response to question 3 clearly shows how homesteads, especially updated ones, have changed over the last decade; I would be pleased for a similar figure, which includes the changes of unchanged homesteads, demolished homesteads, and updated or category-changed homesteads as time changes, to be shown in the figure,to enhance the presentation of the results.

3. Thank you for your textual explanation of the sub-figures in Figure 6. However, I would like you to use different symbols to separately identify the internal intersections, middle intersections, and external intersections, to further enhance the explanatory power of this figure.

4. Thanks for your supplementary description of the categories in Figure 10. However, I am still confused about the process of producing the truth labels in the Between-category Accuracy part. Could you explain how the truth labels are obtained, whether manually identified or based on existing data?

Author Response

We can't upload pictures here. Please see the attachment. Thank you！

Author Response File: Author Response.docx

Reviewer 4 Report

The author has made a lot of revisions as required. It is still recommended that the author check the full text in detail and improve some inappropriate expressions, such as part of "2.2 Data Source"; Line664: "M.K. Hu in 1962"; Line 702: "Wang Xinsheng et al." etc.

In addition, it is recommended that the author organize the discussion content logically and shorten the corresponding content.

Finally, it is strongly recommended that authors do not use the revision mode when revising the article. Please mark the modified content in red or other bright colors. The current version is very unclear and difficult to read.

Minor editing of English language required

Author Response

The format here is easy to be confused. Please see the attachment. Thank you！

Author Response File: Author Response.docx