Salient Object Detection via Fusion of Multi-Visual Perception

Round 1

Reviewer 1 Report

Summary: In this work, the author proposes an effective method for salient object detection by using multi-visual perception following the human visual system, which can quickly identify and focus on impressive objects or areas in complex natural scenarios. Firstly, the corresponding feature map is obtained from the original image. Following that, the salient object detection results of each perception feature are acquired, and then the saliency map is acquired through a feature fusion strategy. Finally, to accurately extract the salient object, superpixel segmentation is employed to remove the interference area. Therefore, this method is a multi-feature approach for salient object detection, where different features complement each other to adapt to complex scenarios. Competitive experiments on challenging datasets demonstrate that the proposed method can effectively detect salient objects in complex natural scenarios. I read the article, and this paper is well-written and contributes to the body of literature. The contents of this paper can be useful for researchers working in this field (e.g., Salient Object Detection). Below, I provide some comments to improve this work further. I believe the following comments can further enhance this paper's quality.

1-      In the abstract, please add the research gap and improvements with numbers. The word effective is a qualitative term and does not justify the efficacy of the proposed method.

2-      In the introduction, please include one more bullet to highlight the experimental details. Please add the details of the datasets used and performance metrics employed to measure performance.

3-      Please define the perceptual features first before use in section # 3. Also, please make the subsection’s headings consistent with Figure 1.

4-      In Figure 5, a,b, and c, please add the axis labels.

5-      In algorithm 1, please add the meaningful cpation, and provide the return (output) statements.

6-      The baselines are mentioned at two different places and they are inconsistent. On line # 351, Ref #: 2 is not used as a baseline while it is a baseline in line # 367. Please rectify these inconsistencies.

7-      Some more discussion about the result before the conclusion is desirable. Authors can provide some details of experiments and challenges that can stem from working on this kind of problem. Also, please discuss the results and why they are like that. This paper lacks experiment-based discussion and therefore, the results cannot be reproduced. Lastly, I suggest discussing the impact of varying the values of  given in Table 1.

8-      What is the uniqueness and novelty of this paper compared to existing work? In my opinion,  many such methods have already been proposed with extensive evaluations, and therefore, there is very little novelty in this work.

9-      Can the proposed approach work with low-resolution or overlapped images? Please add a note in the text.

10-  In the related work section, please provide a summary of the limitations of the existing papers to distinguish the contribution of this work.

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Salient Object Detection via Fusion of Multi-visual Perception” (ID: applsci-2944292). It is our pleasure to receive the very valuable and helpful comments for improving our paper, as well as the precious guide to our research. After studied the comments carefully, we have made correction and response in our new revised manuscript. Revised portion are marked in color in the paper and we hope to meet with approval.The specific comments on the review are provided in the attached PDF.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors present a new method for object detection. It is well-written, and the information is well-structured. However, there is some information, that should be clarified. There is a sentence: "In the MATLAB 2019 environment, running on a CPU with an AMD Ryzen 5 5600X (6 cores), the algorithm processes an image of size 300×400 pixels in an average time of 2.1 seconds." but before that, authors says that "All experiments were run with an AMD Ryzen 7 5800X 8-Core and 32 GB RAM on a 307 Windows 11 operating system. And our method was encoded with MATLAB R2018B. " In this case, please present the average time of the system during all experiments.

Figure 1 is too small. It would also be good to do a deeper analysis of the results obtained.

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Salient Object Detection via Fusion of Multi-visual Perception” (ID: applsci-2944292). It is our pleasure to receive the very valuable and helpful comments for improving our paper, as well as the precious guide to our research. After studied the comments carefully, we have made correction and response in our new revised manuscript. Revised portion are marked in color in the paper and we hope to meet with approval.The specific comments on the review are provided in the attached PDF.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors have presented an interesting paper on the topic of salient object detection. However, the paper could benefit from the following observations:

 - Each abbreviation should be written in full the first time that they are used: MSCN, CNN, FCN.

- Having a citation at the beginning of the phrase is not good practice (line 16).

- Even if statements are usually accompanied by citations, it would help to have some values, graphics to sustain the idea. For example, the statement from line 55, would benefit from having the value needed to train the system. Also, what is the time needed to process an image by a top-down approach vs the authors' proposed solution?

- There is no citation or exemplification of successful examples of salient object detection for the statement from line 102.

- The authors should justify what they mean by "fast and straightforward detection of salient objects" (line 114).

- It is not clear what hardware or setup was used for testing. At line 307 a setup intended for "all experiments" is mentioned, and at line 383, another setup is used. What is the reason for the setup change?

 - The paper would benefit from having a time estimation for the other algorithms used to compare the proposed method. The authors are only mentioning that the proposed technique has an average time of 2.1 seconds, but without mentioning a time value for other algorithms or techniques.

 - Citations 24 and 25 can be improved. With a simple search on the internet, the proper citation for the two articles (author name, journal, volume, issue, etc.) can be found. These two references are quite old (1964 and 1988); would it be possible to search for or use newer research?

The paper would benefit from an English editing service. For example, a sentence should not start with "And" (line 18). The significance has already been emphasized as "important theoretical research significance" (line 33).

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Salient Object Detection via Fusion of Multi-visual Perception” (ID: applsci-2944292). It is our pleasure to receive the very valuable and helpful comments for improving our paper, as well as the precious guide to our research. After studied the comments carefully, we have made correction and response in our new revised manuscript. Revised portion are marked in color in the paper and we hope to meet with approval.The specific comments on the review are provided in the attached PDF.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I have carefully reviewed the revised work and the authors' responses. The authors have reflected on the initial review's feedback and made nice amendments. Below, I provide some minor comments to improve this work further.

1-      In algorithm 1, please add the meaningful cpation, and provide the return (output) statements. The first part of this comment needs to be done again. Please expand the caption as one word seems unsatisfactory.

2-      Please add a reference to algorithm1 from the text and describe its main tasks using line # information.

3-      Please add some more salient keywords for better discovery of this work after publication.

Author Response

Dear Reviewers:

Thank you very much for your letter and for the reviewers’ comments concerning our manuscript entitled “Salient Object Detection via Fusion of Multi-visual Perception” (ID: applsci-2944292). It is our pleasure to receive the very valuable and helpful comments for improving our paper, as well as the precious guide to our research. After studied the comments carefully, we have made correction and response in our new revised manuscript. The detailed response to your review comments is in the attached PDF file.

Author Response File: Author Response.pdf