Research Progress on the Aesthetic Quality Assessment of Complex Layout Images Based on Deep Learning

Round 1

Reviewer 1 Report

A Survey on Image Aesthetic Assessment Abbas Anwar∗ , Saira Kanwal∗ , Muhammad Tahir, Muhammad Saqib, Muhammad Uzair, Mohammad Khalid Imam Rahmani, Habib Ullah

 

Authors have taken a good initiate by considering this problem definition. Yet it would be more appreciable that if they would have discussed the impact of the work too.

Authors have mentioned about aestic grade. It would be good enough if the authors might provide more insights into it.

Refer a few works which might be related to your work:

DOI:10.1007/978-3-030-66519-7_5

 10.3390/jimaging9020030

 

A Survey on Image Aesthetic Assessment Abbas Anwar∗ , Saira Kanwal∗ , Muhammad Tahir, Muhammad Saqib, Muhammad Uzair, Mohammad Khalid Imam Rahmani, Habib Ullah

Minor editing is required.

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a review of existing literature on layout and image aesthetics. The authors also propose to come up with their own method for complex layout images. The information on comparison between existing datasets is present however this cannot be considered a complete or thorough study.

1.      This is a confusing write-up in terms of a review paper that compares and proposes to use layout and aesthetic based studies for complex documents. However, does not compare the methods qualitatively or quantitatively. Comparison between performance of different models is absent which is a critical part of review. Besides comparing different methods their performance in quantitative terms also needs to be compared.

2.       In the abstract, the authors state to propose their own method for the proposed task however there is no clarity in the paper regarding this. There are no results or network diagrams for the proposed approach.

3.       The article proposes to build a dataset for the proposed problem however does not build one.

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper entitled “Research progress on the aesthetic quality assessment of complex layout images based on deep learning” is the topic of the day and is an interesting topic for the reader Also paper was well-structured.

It is recommended to use more recent sources, especially the 2021-2022 and 2023 articles.

It is recommended to add to the content of section 2.3.1 and 2.3.2.

It would be better to compare the results of different methods mentioned in the text in another table similar to Table 1 and 2

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.pdf

Reviewer 4 Report

Manuscript with clear and well described methods for aesthetic quality assessment on complex layout imagese.

Author Response

Dear Editors and reviewers,
Sincerest thanks for the careful and constructive evaluation of our manuscript entitled “Research progress on the aesthetic quality assessment of complex layout images based on deep learning” (ID: applsci-2470347) by the respected reviewers and Editors.   Those evaluations are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches.   We have studied comments carefully and have made correction which we hope meet with approval.   All the changes are highlighted with red color throughout the manuscript.

OK.

We deeply appreciate your consideration of our manuscript. If you have any queries, please don’t hesitate to contact me at the address below.

Thank you and best regards.

Yours sincerely,

Prof. Yunfei Zhong

Corresponding author:

Name: Prof. Yunfei Zhong

E-mail: [email protected]

Reviewer 5 Report

In this paper, based on complex layout images such as posters, covers, and brochures, we first analyze complex layout images, review traditional and deep learning-based layout image analysis methods and aesthetic quality assessment methods; and propose a method for aesthetic quality assessment of images based on deep learning combined with complex layout analysis. More detailed comments are given as follows:

 

 1-   The abstract need to rewrite.

2-   In abstract, the author mention that to proposed ….., but I don’t see any proposed method.

3-   In 2.2. Traditional layout analysis methods section, it better to write it as table to show the power ,advantage, disadvantage, different, gab points.

4-   Figure 2 need to redraw.

5-   2.3.1.. Layout analysis method based on Support Vector Machine (SVM) (as point 3).

6-   2.3.2. Layout analysis based on Neural Networks (as point 3).

7-   Where the block diagram of the proposed system.

8-   Discuss the limitations of the proposed method.

 

no need, it is ok

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.pdf

Reviewer 6 Report

Dear Authors,

   The article: "Research progress on the aesthetic quality assessment of complex layout images based on deep learning" presented by you presents an interesting topic with many challenges to overcome. I have provided comments on the digital file and would like your attention to the following suggestions:

1) Chapter 1: preprocessing which in some cases encompasses different methods among them segmentation has different computational costs. It would be interesting to mention this issue and discuss it in the following chapters. Present the inherent advantages of employing automated methods for layout analysis, as well as, the copyright issue can be benefited by using these methods.

2) Chapter 2.2: Present the most commonly employed segmentation method. Currently, different methods are available and usually, the computational cost varies from method to method.

3) Chapter 2.3: The SVM algorithm has some limitations regarding the proper choice of parameters and the appropriate kernel. These issues should be better explored and presented to the reader.

4) Chapter 2.3: "SVM-based image segmentation method". SVM is actually a classification algorithm. Segmentation can be achieved with other algorithms. Please review.

5) Chapter 4: Table 2 introduces the research performed with their respective score. But it would be interesting to discuss the presented studies with their advantages and disadvantages. Computational cost, accuracy?

  The chapter 4 should present a deeper discussion highlighting the contributions of each of these methods and their applicability.

   I conclude my review by congratulating you on the bibliographic research performed and on the presented version of the manuscript.

 

Respectfully,

Comments for author File: Comments.pdf

Review the use of punctuation, especially the use of the comma. In some sections, long paragraphs are presented and this tends to make reading and consequently understanding difficult. When presenting acronyms for the first time in the text, present their meaning.

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have included new comparison tables and tried to address my comments.

Usually, when the authors propose an approach/method it is backed by experimental findings and compared with the SOTA in terms of performance. I understand that there is no dataset at this point and thus experiments are not possible. Proposing a method is probably fine when dataset is ready to work with since then it can be backed by experimental findings.

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.docx

Reviewer 6 Report

Dear Authors,

   The article entitled: "Research progress on the aesthetic quality assessment of complex layout images based on deep learning" presents an interesting topic that explores the analysis of layouts from a human/software point of view. The topic is relevant and presents challenges to be overcome despite advances in algorithms, especially those based on deep learning.

   I have read the second version of the manuscript very carefully and using my observations from the first stage of revision supported by the cover letter I have found that all my suggestions have been duly implemented or justified.

    I congratulate you on the final version of the manuscript and thank you for sending the cover letter.

Respectfully,

Specific adjustments regarding the use of commas are still necessary.

Author Response

Thank you very much for your comments, Please see the attachment for details.

Author Response File: Author Response.docx