Active Visual Perception Enhancement Method Based on Deep Reinforcement Learning

Round 1

Reviewer 1 Report

To effectively detect low confidence targets，the paper introduces a visual perception enhancement method leveraging deep reinforcement learning, enabling PTZ cameras to achieve active vision, optimizing object detection and information acquisition, and enhancing their intelligent perception capabilities. However, there are some suggestions and problems in the paper as follows:

1. The format of references is messy and not uniform; there are more earlier literature on arXiv, and the published articles should be cited.

2. Lack of comparative experiments with existing algorithms, it is suggested to supplement the comparative experiments in the last three years.

3. Why choose the yolo3 network instead of the popular yolo5-8 updated version?

4. The conclusion is too long, and it is suggested to streamline it.

5. There are more grammatical errors in the paper, such as “Jin et al. [17] use object detection models....” it should be the past tense.

6. The Related work section is not well researched.

7. English needs to be improved.

8. Innovation is a bit lacking.

English needs to be improved.

Author Response

Dear reviewer:

Thank you for your decision and constructive comments on my manuscript. We have carefully considered the suggestion of Reviewer and make some changes. We have tried our best to improve and made some changes in the manuscript.

Please see the attachment

Thanks for all the help.

Best wishes

Zhonglin Yang

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper proposes a visual perception enhancement method based on deep reinforcement learning. Here, I have a few concerns:

1. The paper is in general well organized and easy to read. But the motivation should be explained in more detail.

2. Are there any limitations to the methods proposed in the paper? 

3. The Action part of figure2 is unreadable.You could pay attention to details of your figures.

4. I suggest you provide a detailed introduction of your model.

5. The paper requires significant improvement in its English language usage.

6. The conclusion section can be written in more detail.

7. It is better for the authors to review the important related works, such as doi: 10.1016/j.eswa.2023.122013, to facilitate the potential readers.

 The paper requires significant improvement in its English language usage.

Author Response

Dear reviewer:

Thank you for your decision and constructive comments on my manuscript. We have carefully considered the suggestion of Reviewer and make some changes. We have tried our best to improve and made some changes in the manuscript.

Please see the attachment

Thanks for all the help.

Best wishes

Zhonglin Yang

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors present a proposal for active vision in which the PZT camera can make decisions that improve the perception of the environment. To do this, an integrated reward function is used that optimizes results. The evaluation of this proposal has been carried out using a simulation environment implemented ad-hoc.

The work is well structured. The authors present sufficient related bibliography and make clear the main contributions of the work. Likewise, they describe the main weaknesses of the proposed method, such as the design of the state space to introduce more characteristics.

Line 138: format of W

Line190: format of [Sx,Sy]

Line 202: redundant "multiplied by the multiplication"

Line 217:  "are the reward weights"

Lines 282-287 and 292-296: format paragraphs

Line 327: format  e=0.1

Line 349: Figure 5 is referenced instead of Figure 6.

Author Response

Dear reviewer:

Thank you for your decision and constructive comments on my manuscript. We have carefully considered the suggestion of Reviewer and make some changes. We have tried our best to improve and made some changes in the manuscript.

Please see the attachment

Thanks for all the help.

Best wishes

Zhonglin Yang

Author Response File: Author Response.pdf

Reviewer 4 Report

The article deals with a topic already studied previously but where it is intended to offer improvements. In this sense, it is very important that the authors show the limitations that exist in the state of the art. This aspect must be emphasized. The state of the art section is very brief compared to the introduction. It is suggested to rewrite these sections in a more appropriate way. Some figures do not clearly show the details they want to show. It is suggested to modify them to emphasize those aspects that you want to emphasize. Table 4 asks to justify the success rates that only vary in steps of 10%. The summary provides a result of 22.90% that does not appear in the article. The information in summary and conclusions must be consistent with what is presented in the rest of the paper. The format of references should be reviewed.

Author Response

Dear reviewer:

Thank you for your decision and constructive comments on my manuscript. We have carefully considered the suggestion of Reviewer and make some changes. We have tried our best to improve and made some changes in the manuscript.

Please see the attachment

Thanks for all the help.

Best wishes

Zhonglin Yang

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

My previous concerns have been considered