Late Deep Fusion of Color Spaces to Enhance Finger Photo Presentation Attack Detection in Smartphones
Round 1
Reviewer 1 Report
Some kind of cross validation should be used for the experiments.
A statistical test should be used for the comparison of the examined methods.
The authors should better explain why the proposed methodology seems to work well and present some information about the time efficiency of their method.
Author Response
Reviewer 1:
1. Some kind of cross-validation should be used for the experiments.
In Table 3, we added a column that reports cross-validation results for every model considered in this study.
2. A statistical test should be used for the comparison of the examined methods.
At the end of the proposed approach, we described the statistical test used for comparing classification methods and their interpretation. At the end of the results section, we described our experimental findings.
3. The authors should better explain why the proposed methodology seems to work well and present some information about the time efficiency of their method.
In the results section, we mention the time efficiency of the models used. Table 1 shows the CNN models used in our work and their floating-point operations. We also elaborated our proposed methodology and added additional statistical analysis to strengthen our proposed method.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors proposed the PAD attack detection framework for smartphone security. It's a good manuscript, but many flaws have to be fixed.
- Title should be more specific and to be relevant to your research topic.
- Some sentences should be revised/refined. E.g., “A finger photo is a fingerprint image acquired using a basic smartphone camera.” The explanation of the term should be improved. It would be better if you introduce the spoofed image, rather than adding the explanation of the finger photo.
- The authors needed to double-check the typos in both the sentences and the figures.
- The style, “[? ]” is difficult to track the cited works in which items are relevant or not.
- The most important thing is that it is not clear what are the research gaps in your work. How to solve that?
- The explanation should be clearer between the nexus of methodologies in sections 3 and 4.
Author Response
Reviewer 2:
1. The title should be more specific and to be relevant to your research topic.
The title has been edited: it is now more specific by representing a better description of the proposed work.
2. Some sentences should be revised/refined. E.g., “A finger photo is a fingerprint image acquired using a basic smartphone camera.” The explanation of the term should be improved. It would be better if you introduce the spoofed image, rather than adding the explanation of the finger photo.
This definition has been improved. A background of the technology and details about spoofed images have been provided as well. [see the Abstract]
3. The authors needed to double-check the typos in both the sentences and the figures.
The manuscript has been proofread for typos.
4. The style, “[? ]” is difficult to track the cited works in which items are relevant or not.
The issue has been solved.
5. The most important thing is that it is not clear what the research gaps are in your work. How to solve that?
The research gaps have been clarified in the introduction section before describing the contribution of the paper.
6. The explanation should be clearer between the nexus of methodologies in sections 3 and 4.
The explanation of the link between section 3 and 4 has been improved.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comment 1: The introduction needs to be organized. Moreover, some key dimensions of the paper are missing in the section. It should cover originality of the work along with paper organization.
Comment 2: In the introduction section of the paper, the authors must justify why the defense against finger photo spoofing is significant. Already there are many defense systems, then how your approach is different from those already existing? In addition, the paper layout is lacking. Authors need to highlight the research contribution
Comment 3:In terms of the performance of late fusion on the overall system, it can be seen that Interpretation of test results shown in figure 6 is not satisfactory. Discuss the significance of the results obtained in detail. Moreover, SECTION 5 is not presenting worthy outputs of the findings. Improve 5.1 and 5.2 in the revised manuscript.
Comment 4: ALL the figures BEFORE conclusion section.
Comment 5:
Not citing sources properly. Use a cross-reference.
Comment 6:The authors did NOT formulate ANY research questions (RQ) in the paper. In other words, before jumping directly to the RESEARCH METHOD/STUDY DESIGN, the authors are strongly recommended to start with the activity that aims at formulating RQs based on the objectives of the research in order to FOCUS their research. For example, the authors should write: The following (two/three/....) research questions have motivated the work reported in this paper:
RQ1: How can the robustness of the PAD system be improved using deep fusion of the best color models?
Comment 7: The authors need to go through the Keywords one more time to make them more CONCISE. They might need to re-select them in order to represent the whole research and cover the important parts of this study.
Comment 8 :Important comments about (Spelling, Style, and Grammar) mistakes:The logical structure of this paper is good. Try to improve the usage of English grammar.In this paper, there are many "we", [The word “we” is repeated MORE than 19 times ]. Don't keep repeating the word “we” over and over. It is very important to avoid the use of “we”, and it is recommended to use passive voice or present sentences.
There are some overlapped sentences in the manuscript; authors are suggested to check and modify the manuscript.
Author Response
Reviewer 3:
1. The introduction needs to be organized. Moreover, some key dimensions of the paper are missing in the section. It should cover originality of the work along with paper organization.
This Section has been restructured. Now it includes a clear statement of this paper’s contribution as well as a description of the motivation of the proposed work.
2. In the introduction section of the paper, the authors must justify why the defense against finger photo spoofing is significant. Already there are many defense systems, then how is your approach different from those already existing? In addition, the paper layout is lacking. Authors need to highlight the research contribution.
We have clarified the research contribution of this paper in the introduction section. We also have discussed how the proposed approach differs from the existing methods.
3. In terms of the performance of late fusion on the overall system, it can be seen that the Interpretation of test results shown in figure 6 is not satisfactory. Discuss the significance of the results obtained in detail. Moreover, SECTION 5 is not presenting worthy outputs of the findings. Improve 5.1 and 5.2 in the revised manuscript.
Section 5 has been rewritten.
4. ALL the figures BEFORE the conclusion section.
The figures and tables are now all before the conclusion section.
5. Not citing sources properly. Use a cross-reference.
The format of the citations has been corrected.
6. The authors did NOT formulate ANY research questions (RQ) in the paper. In other words, before jumping directly to the RESEARCH METHOD/STUDY DESIGN, the authors are strongly recommended to start with the activity that aims at formulating RQs based on the objectives of the research in order to FOCUS their research. For example, the authors
should write: The following (two/three/....) research questions have motivated the work reported in this paper:
RQ1: How can the robustness of the PAD system be improved using a deep fusion of the best color models?
The two research questions under investigation in this paper have been defined and contextualized in the introduction.
7. The authors need to go through the Keywords one more time to make them more CONCISE. They might need to re-select them in order to represent the whole research and cover the important parts of this study.
The keywords have been revised.
8. Important comments about (Spelling, Style, and Grammar) mistakes: The logical structure of this paper is good. Try to improve the usage of English grammar. In this paper, there are many "we", [The word “we” is repeated MORE than 19 times]. Don't keep repeating the word “we” over and over. It is very important to avoid the use of “we”, and it is recommended to use passive voice or present sentences.
The expressions with “we” have been significantly reduced.
9. There are some overlapped sentences in the manuscript; authors are suggested to check and modify the manuscript.
Overlapping sentences have been identified, removed and rewritten.
Author Response File: Author Response.pdf
Reviewer 4 Report
In this paper, the deep fusion method for detecting spoofing finger photo is mentioned. The idea is interesting. However, the result is not so much improved compared with the single detector using single color space. It may not be meet to the additional computational cost.
This paper is not enough edited. All citations are [??], and from page 7 to 8 some sentences are considered to be lacked.
This paper should be rewritten with more outperforming experimental result.
Author Response
1. This paper is not enough edited. All citations are [??], and from page 7 to 8 some sentences are considered to be lacked.
The citations have been formatted correctly, and overlapping sentences have been eliminated and redone. The section has also been reorganized. The contribution of this paper is now stated clearly, and the motivation for the suggested work is explained
2. This paper should be rewritten with the more outperforming experimental result.
To prove our methodology, we introduced statistical results and also showed additional cross-validation results.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper could be accepted in the current form
Author Response
We proofread the paper. English language typos and spelling mistakes were removed during proofreading.
Reviewer 2 Report
Thanks for your revision, improving with the comments on the previous version. However, some typos and grammar issues are found. e.g., ".......MobileNetV3, and DenseNet-121 [26–28,28,29]". Proofing reading is still necessary for the final one.
Author Response
We proofread the paper, and errors like those mentioned and minor spelling mistakes were removed.
Reviewer 4 Report
The contents of this paper are well edited. However I have some comments.
In abstract, finger photo recognition is mentioned as a promising touch less technology. However, for smartphones, touches technology is not so much required because users touch smartphone anyway to operate it, and cost of fingerprint sensors is not high compared with the overall cost of smartphone. I think that this method should be used in public spaces, such as bank, entrance of building, and so on, where many people must touch fingerprint sensor for authentication, or on devices which does not generally have fingerprint sensors, such that low cost PC, gaming consoles, and iPhone using Face ID when the mask is required to avoid infection.
In table 3, the results of state of art methods are added. As mentioned in paper, AlexNet[16] outperformed proposed method. If the condition of the experiments is completely different as mentioned in this revision of paper, this result should be removed to avoid the confusion of reader.
Author Response
Q1: In abstract, finger photo recognition is mentioned as a promising touch less technology. However, for smartphones, touches technology is not so much required because users touch smartphone anyway to operate it, and the cost of fingerprint sensors is not high compared with the overall cost of smartphone. I think that this method should be used in public spaces, such as bank, entrance of building, and so on, where many people must touch fingerprint sensor for authentication, or on devices which does not generally have fingerprint sensors, such that low-cost PC, gaming consoles, and iPhone using Face ID when the mask is required to avoid infection.
In the abstract, the sentence is elaborated as follows:
Finger photo recognition represents a promising touchless technology that offers smartphones portable and hygienic authentication solutions, eliminating physical contact. Public spaces, such as banks and staff-less stores, benefit from contactless authentication considering the current public health sphere.
2. In table 3, the results of state of the art methods are added. As mentioned in the paper, AlexNet[16] outperformed proposed method. If the condition of the experiments is completely different as mentioned in this revision of paper, this result should be removed to avoid the confusion of reader.
To avoid confusion, we removed AlexNet[16] experiment.
Author Response File: Author Response.pdf
