Generalized Image Captioning for Multilingual Support

Round 1

Reviewer 1 Report

The paper presents the evaluation of some image caption and text generation methods.

The research paper still needs to be completed, and I advise improving it through the following additions, improvements, and corrections.

Abstract: 90% of the abstract is dedicated to introducing and justifying the approached paper problem. Just the last sentence describes what the objective of the paper is. I advise rewriting the abstract and dividing this into three small blocks: (1) the introduction and justification of the approached problem, (2) the method, technique, or algorithm (or study) proposed by the research, and (3) the final depiction of the found results and evaluation of these results.

Line 60: OCR is not the acronym for Optical Recognition Software. This acronym is an example of that kind of software. This sentence should be modified.

Before section 3, the paper should include a paragraph listing and explaining the paper's contributions. This contribution should be related to the topics addressed in the work review depicted previously.

Line 176: The sentence needs to be clarified. Please rewrite this sentence.

Table 2 and Table 3: The technique's name is wrong in the table's caption. Belu instead of Bleu

Table 4: the second description of the "Filter Captioning Model" is wrong

Line 295: the references of Figures and table are missing

The conclusion paragraph should be rewritten. In addition, some sentences need to be clarified. Please change the sentence of line 315, removing "You" and approaching a third-person narrative.

In general, the paper should focus more on the contributions made to state of art and emphasize these contributions, in addition to carrying out a deeper comparative study of the proposed method and the methods found in the state-of-the-art.

An analysis of the pros and cons of the proposed method, as well as its limitations, should also be included.

Author Response

Thank you for your review of the paper.

- I modified the OCR, BLEU, Table Caption, etc. you gave me feedback.
- We have revised the last sentence of the conclusion that adds the limitations of the proposed method.
- Abstract has been modified more explicitly.
- Introduction has modified the distribution more explicitly.
In addition, minor errors such as English grammar have been corrected, and I am planning to apply for the English edit service.

Thank you again for your review.

Author Response File: Author Response.pdf

Reviewer 2 Report

The issue of automation of descriptions of graphics, photos, frames from the film is extremely important from the point of view of facilitation for people with disabilities. You have taken up such a challenge, and the proposed method makes sense and the results confirm this. The description shows that the bases were natural images. What would the results look like for abstract images (e.g., art in pictures)? Wouldn't domain errors appear ?

Author Response

Thank you for your review of the paper.

I corrected the typo and corrected the last sentence of the conclusion that added the limitations of the method of suggestion.
Abstract has been modified more explicitly, and Introduction has modified the content more explicitly.
In addition, minor errors such as English grammar have been corrected, and I am planning to apply for the English edit service.

Thank you again for your review.

Author Response File: Author Response.pdf

Reviewer 3 Report

Summary: This paper evaluates commonly available Korean and English image caption models and text generation techniques suitable for image caption purposes

 

1. This paper lacks of scientificity.

2. The writing style of this paper should be enhanced.

3. I cannot clearing understand the main contribution of this paper. It is suggested that the authors should list the main contribution of this paper in a separate paragraph.

4. The references are obsoleted. The authors should add more recent references to review.

5. In the experiments, could you please provide more comparisons with more recent methods?

6. Fig.8-11 should be improved and enhanced. The provided figures are so ugly. Please provide good form of the figures.

7. Could you provide the failure case of the proposed method?

8. The writing style of the abstract should be improved. In the abstract part, the reviewer cannot understand the core contribution and method.

9. Please explain the obvious differences with previous approaches.

Author Response

Thank you for your review of the paper.

- Abstract has been modified more explicitly.
- Introduction has modified the distribution more explicitly.
- Reflecting the feedback you gave me, I edited Figure 8, Figure 9, Figure 10, and Figure 11 into one picture.
- We have revised the last sentence of the conclusion that adds the limitations of the proposed method.
- Corrected typos present in the paper.
In addition, minor errors such as English grammar have been corrected, and I am planning to apply for the English edit service.

Thank you again for your review.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The provided response letter has addressed major concerns.