Restoration of Dimensions for Ancient Drawing Recognition

Round 1

Reviewer 1 Report

The paper studied calculations of the actual dimensions used in the drawings of Tonga and Eonjo contained in the Jaseungcha Dohae written by Gunnam Ha BaeckWon. Using this conversion, the actual dimensions used in year 1800s were restored, specifically for the southern region of Korea. Through this study, the paper introduced new data for the analysis of ancient drawings in the field of drawing recognition and analysis. 

The paper is overall well written and well organized. However, the paper needs thorough revisions to improve the quality of paper as follows:

- The significance of the paper is not clear. It is suggested to provide more details about the objectives and significance of the proposed work.
- Some background of the proposed work and related work could be added to the paper. This is important to the readers who might not have the background of this research, particularly Korean history.
- CNN part could be updated if relevant.
- The tables are not consistent. (Sizes and fonts)
- Some minor typos.
 
 

Author Response

The paper studied calculations of the actual dimensions used in the drawings of Tonga and Eonjo contained in the Jaseungcha Dohae written by Gunnam Ha BaeckWon. Using this conversion, the actual dimensions used in year 1800s were restored, specifically for the southern region of Korea. Through this study, the paper introduced new data for the analysis of ancient drawings in the field of drawing recognition and analysis. 

The paper is overall well written and well organized. However, the paper needs thorough revisions to improve the quality of paper as follows:

- The significance of the paper is not clear. It is suggested to provide more details about the objectives and significance of the proposed work.

For this reason, classical drawing recognition is difficult due to the presence of many different dimensions. This paper has the meaning of restoring different metrology lengths using proportional expressions.

- Some background of the proposed work and related work could be added to the paper. This is important to the readers who might not have the background of this research, particularly Korean history.

I supplemented it through CNN update. [16]

- CNN part could be updated if relevant.

[16] Gyllenbok, Jan, Jan Gyllenbok, and Goob. Encyclopaedia of historical metrology, weights, and measures. Cham: Birkhäuser, 2018.

- The tables are not consistent. (Sizes and fonts)

Modified the size and font of the table.

- Some minor typos.

I reviewed the typos and grammar again.

Author Response File: Author Response.pdf

Reviewer 2 Report

This article tries to investigate and determine the actual size of the “cheok” scale, the traditional weights and measures of Korea, and tries to complete a part of data construction on the recognition of ancient drawings in the field of artificial intelligence.

But there are some problems as follows:

1、In the introduction part, there are too many descriptions of the inconsistency of size, historical quotations and background. The authors can introduce some related work about how other people deal with it, and what are their methods’ advantages and weaknesses.

2、Maybe it’s better to clearly display your contributions and innovations.

3、It is recommended to use a flowchart to describe how to convert the size in the article used.

4、 At the beginning of the article, the authors say the technology primarily based on convolutional neural networks, but we don’t see how they use it. The method part should be more detailed, don't focus on describing the history of varying sizes.

5、The purpose mentioned in the abstract is to complete data construction for artificial intelligence. It is best to introduce how to use artificial intelligence to analyze or use the data.

6、 The article represents many results, but it’s a little difficult for us to believe in those results because the authors don’t describe enough details about the implementation process.

7、Some of the charts can be modified in format and layout, which will make them more beautiful and easy to understand.

8、Although most of the references are rather new, there are not enough references for people in other fields to understand the research.

Comments for author File: Comments.pdf

Author Response

This article tries to investigate and determine the actual size of the “cheok” scale, the traditional weights and measures of Korea, and tries to complete a part of data construction on the recognition of ancient drawings in the field of artificial intelligence.

But there are some problems as follows:

1、In the introduction part, there are too many descriptions of the inconsistency of size, historical quotations and background. The authors can introduce some related work about how other people deal with it, and what are their methods’ advantages and weaknesses.

I modified the contents of the introduction.

2、Maybe it’s better to clearly display your contributions and innovations.

For this reason, classical drawing recognition is difficult due to the presence of many different dimensions. This paper has the meaning of restoring different metrology lengths using proportional expressions.

3、It is recommended to use a flowchart to describe how to convert the size in the article used.

The following paper will cover classical drawing implementation, so we will cover detailed algorithms in the next paper.

4、 At the beginning of the article, the authors say the technology primarily based on convolutional neural networks, but we don’t see how they use it. The method part should be more detailed, don't focus on describing the history of varying sizes.

I modified the history of size in the content of abstract.

5、The purpose mentioned in the abstract is to complete data construction for artificial intelligence. It is best to introduce how to use artificial intelligence to analyze or use the data.

Part of the introduction has been modified and supplemented

6、 The article represents many results, but it’s a little difficult for us to believe in those results because the authors don’t describe enough details about the implementation process.

Top-up content in Chapter 3

Due to the lack of measurement data on the classical drawings, the proportional formula was measured on the data described. Measurement methods and analysis follow the following sequence:
- Collection of drawings describing proportional formulas
-Copy classical drawings to same size
- Measure the length of each line in the drawing
- Store neural network learning data for each line segment
- Machine learning is implemented through the average value of each line separation.

7、Some of the charts can be modified in format and layout, which will make them more beautiful and easy to understand.

Modified to match format and layout.

8、Although most of the references are rather new, there are not enough references for people in other fields to understand the research.

References have been supplemented [16]

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper addresses the challenge of restoring the actual scales and measurements used and applied in Korea in 1800s with drawings presented in Jaseungcha Dohae. Based on this study, new data for the analysis of ancient drawings was obtained in field of drawing recognition and analysis. Overall, the submission present an interesting research work, which may guide further research on the related topic but also reveal methods applicable to modern measurement and scaling. Despite of this, the authors should put additional effort and present their research method and the applicability of both method and outcomes, to modern research actions. This reviewer suggests to address the following comments:

• The introduction need to improve the presentation of the core highlights of the conducted research. They should be clearly enumerated emphasizing their differences against state-of-the-art work.
• The manuscript addresses a very specific topic, which results are clearly relevant to the researchers on Jaseungcha Dohae and attempts to transpose its measurement to the modern context. Since this may be an small part of the average journal audience, It is necessary to discuss how the research outcomes are applicable on the modern age and what benefices they serve to the ongoing advantages to the topics targeted by this journal.
• Can the adopted method be applied to other related studies?. In this case, it may be clarifying to explicitly present its design principles: primary/secondary goals, null/alternative research hypothesizes operational assumptions (premises), assumed limitations/constraints, etc.
• The paper lacks of an analytical section (e.g. “Discussions” section) that comments and discuss the research method, outcomes, and correlated them against related state of the art efforts.

Author Response

The paper addresses the challenge of restoring the actual scales and measurements used and applied in Korea in 1800s with drawings presented in Jaseungcha Dohae. Based on this study, new data for the analysis of ancient drawings was obtained in field of drawing recognition and analysis. Overall, the submission present an interesting research work, which may guide further research on the related topic but also reveal methods applicable to modern measurement and scaling. Despite of this, the authors should put additional effort and present their research method and the applicability of both method and outcomes, to modern research actions. This reviewer suggests to address the following comments:

• The introduction need to improve the presentation of the core highlights of the conducted research. They should be clearly enumerated emphasizing their differences against state-of-the-art work.

Part of the introduction has been modified and supplemented

• The manuscript addresses a very specific topic, which results are clearly relevant to the researchers on Jaseungcha Dohae and attempts to transpose its measurement to the modern context. Since this may be an small part of the average journal audience, It is necessary to discuss how the research outcomes are applicable on the modern age and what benefices they serve to the ongoing advantages to the topics targeted by this journal.

1.References have been supplemented

2. For this reason, classical drawing recognition is difficult due to the presence of many different dimensions. This paper has the meaning of restoring different metrology lengths using proportional expressions.

• Can the adopted method be applied to other related studies?. In this case, it may be clarifying to explicitly present its design principles: primary/secondary goals, null/alternative research hypothesizes operational assumptions (premises), assumed limitations/constraints, etc.

Top-up content in Chapter 3

• Due to the lack of measurement data on the classical drawings, the proportional formula was measured on the data described. Measurement methods and analysis follow the following sequence:
  - Collection of drawings describing proportional formulas
  -Copy classical drawings to same size
  - Measure the length of each line in the drawing
  - Store neural network learning data for each line segment
  - Machine learning is implemented through the average value of each line
  
  
• The paper lacks of an analytical section (e.g. “Discussions” section) that comments and discuss the research method, outcomes, and correlated them against related state of the art efforts.

I supplemented the contents in chapters 2 and 3.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

It's obviously that the authors have made some modifications. First, most of the charts have been modified in format and layout, which are more beautiful. And in the Introduction and Abstract part,  background information has been less and  authors' own research has got more attention.

But there are still some problems as follows:

1、This is the same queation I have mentioned in the 1^st round review. The most important thing that confused me is about the AI algorithm. At the beginning of the article, the authors say the technology primarily based on convolutional neural networks, but we don’t see how they use it. The method part should be more detailed. It's better to give us more infomation about the algorithm, the method to train, how to set hyperparameters and so on. Showing whole framework and specific details will improve the reliability and credibility of research

2、There are also some points about the set type:

In the 118th line, there should be a blank space after the dash.

In the 155th line, it's not so suitable to depart the title from the chart. Leaving them in the same page will be better.

In the 177th and 221th lines, there are wired black regions near the Sigma in the formula, which seems to be a mistake while writing the formula?

I think if these problems are solved, the paper will look better.

Comments for author File: Comments.pdf

Author Response

It's obviously that the authors have made some modifications. First, most of the charts have been modified in format and layout, which are more beautiful. And in the Introduction and Abstract part,  background information has been less and  authors' own research has got more attention.

But there are still some problems as follows:

1、This is the same queation I have mentioned in the 1st round review. The most important thing that confused me is about the AI algorithm. At the beginning of the article, the authors say the technology primarily based on convolutional neural networks, but we don’t see how they use it. The method part should be more detailed. It's better to give us more infomation about the algorithm, the method to train, how to set hyperparameters and so on. Showing whole framework and specific details will improve the reliability and credibility of research

I modified flowchart in chapter3. More detailed implementations and experiments will be covered in the next paper.

2、There are also some points about the set type:

2-1. In the 118th line, there should be a blank space after the dash.

I modified it.

2-2. In the 155th line, it's not so suitable to depart the title from the chart. Leaving them in the same page will be better.

I modified it.

2-3. In the 177th and 221th lines, there are wired black regions near the Sigma in the formula, which seems to be a mistake while writing the formula?

I modified it.

Author Response File: Author Response.pdf