GP-Net: Image Manipulation Detection and Localization via Long-Range Modeling and Transformers

Round 1

Reviewer 1 Report

Jin Peng Tran and collaborators presented an innovative approach to image manipulation localization, which they called GPNet.

Reasons for my decision:

- The abstract is adequate.

- The introduction is adequate.

- The methodology was presented adequately.

- The results were presented adequately and the contributions of the work to the study area were very clear.

Conclusion:

Based on my observations, I conclude that the work has good relevance for the proposed area of application and I was unable to identify serious problems in the text, which is why I suggest publishing it as is.

Author Response

Dear Editors and Reviewers:

Thank you very much for reviewing our paper and for your comments and suggestions. I'm glad you approve of our paper.

       We will continue to strive to do a better work in the future and thank you again for your comments and suggestions.

With best wishes,

Yours sincerely.
Jin Peng.

Author Response File: Author Response.pdf

Reviewer 2 Report

Major Points:

Strengths:

• The paper addresses a timely and relevant topic, focusing on the challenges of image manipulation techniques.

• The proposed method, GP-Net, combines CNNs and Transformers, aiming to overcome limitations in existing methods related to long-range modeling.

• The results section provides a clear comparative analysis, showcasing the performance of GP-Net against benchmark methods.

Areas for Improvement:

• The abstract would benefit from a brief mention of the main findings or the impact of the GP-Net method.

• In the methodology section, deeper insights into the workings of GP-Net before the comparative analysis would aid understanding.

• The discussion section is missing. Including a discussion on the implications of the results, comparisons with other studies, and potential real-world applications of GP-Net would offer added depth.

Recommendations:

• Enhance the abstract with a summary of results or the impact of GP-Net.

• Delve deeper into the proposed GP-Net methodology, detailing its architecture and workings.

• Consider adding a discussion section to provide interpretations of the results and comparisons with existing studies.

Final Recommendation: Accept with minor revisions.

Summary: The paper presents valuable contributions to the field of image manipulation detection. With the suggested revisions, it promises to provide a comprehensive understanding of the GP-Net method and its advantage

Author Response

Dear Editors and Reviewers:

       Thank you for your nice comments on our article. According to your suggestions, we have supplemented several data here and corrected several mistakes in our previous draft. Based on your comments, we also attached a point-by-point letter to you . We have made revisions to our previous draft. The detailed point-by-point responses are listed below.

       Thanks for your help. We feel sorry that we did not provide enough information about that. Your suggestion is a greater improvement of our article.

Point1： Enhance the abstract with a summary of results or the impact of GP-Net.

Response1: We have enhanced the abstract with the results achieved by GPNet on a variety of image tampering datasets.

Point2:   Delve deeper into the proposed GP-Net methodology, detailing its architecture and workings.

Response2: We have added a description of the network structure at the beginning of the  Section 3, explaining how the network works based on your comments.

Point3: Consider adding a discussion section to provide interpretations of the results and comparisons with existing studies.

Response3: We add a discussion section in section 4.2 in terms of visualization based on your comments and compare it with existing studies.

We tried our best to improve the manuscript which will not influence the content and framework of the paper. We appreciate for Editors/Reviewers’ warm work earnestly, and hope the correction will meet with approval. Once again, thank you very much for your comments and suggestions.

With best wishes,

Yours sincerely.

Jin Peng.

Author Response File: Author Response.pdf

Reviewer 3 Report

This manuscript proposes GPNet as a new ANN architecture that combines ViT and CNN for detecting forgery in images. In general, this manuscript is well written and addresses one of the most important topics for computer vision. However, it would be better if the evaluation of the comparison included some architectures developed for the same goal as this work.

Author Response

Dear Editors and Reviewers:

Thank you for your nice comments on our article. According to your suggestions, we have supplemented several data here and corrected several mistakes in our previous draft. Based on your comments, we also attached a point-by-point letter to you. We have made revisions to our previous draft. The detailed point-by-point responses are listed below.

Thanks for your help. We feel sorry that we did not provide enough information about that. Your suggestion is a greater improvement of our article.

Point1：It would be better if the evaluation of the comparison included some architectures developed for the same goal as this work.

Response1: We add a discussion section in Section 4.2 and compare it with existing research where Objectformer is also the architecture of CNN and Transformer

       We tried our best to improve the manuscript which will not influence the content and framework of the paper. In this revised version, changes to our manuscript were all highlighted within the document by using red colored text. We appreciate for Editors/Reviewers’ warm work earnestly, and hope the correction will meet with approval. Once again, thank you very much for your comments and suggestions.

With best wishes,

Yours sincerely.
Jin Peng.

Author Response File: Author Response.pdf

Reviewer 4 Report

I advise including the following notes:

*The description of Figure 1 should be reviewed.

*Enhance the content in both section 4.3 and section 4.4 by providing additional information.

*Provide clarification for equation 4 (line 181).

*Enhance the understanding of Figure 4 by providing a more detailed explanation.

*I recommend revising the conclusion to make it more concise and informative.

The article is generally well-written, but some minor proofreading is required.

Author Response

Dear Editors and Reviewers:

       Thank you for your nice comments on our article. According to your suggestions, we have supplemented several data here and corrected several mistakes in our previous draft. Based on your comments, we also attached a point-by-point letter to you and. We have made revisions to our previous draft. The detailed point-by-point responses are listed below.

       Thanks for your help. We feel sorry that we did not provide enough information about that. Your suggestion is a greater improvement of our article.

Point1: The description of Figure 1 should be reviewed.

Response1: We have rewritten the description of Figure 1 based on your comments.

Point2: Enhance the content in both section 4.3 and section 4.4 by providing additional information.

Response2: We have enhanced the content in both section 4.3 and section 4.4.

Point3: Provide clarification for equation 4 (line 181).

Response3: We have changed this mistake.

Point4: Enhance the understanding of Figure 4 by providing a more detailed explanation..

Response4: We add a discussion section in section 4.2 in terms of visualization for Figure 4 based on your comments.

Point5: I recommend revising the conclusion to make it more concise and informative..

Response5: We have rewritten conclusion based on your comments.

We tried our best to improve the manuscript which will not influence the content and framework of the paper. In this revised version, changes to our manuscript were all highlighted within the document by using red colored text. We appreciate for Editors/Reviewers’ warm work earnestly, and hope the correction will meet with approval. Once again, thank you very much for your comments and suggestions.

With best wishes,

Yours sincerely.
Jin Peng.

Author Response File: Author Response.pdf