Leveraging Deep Features Enhance and Semantic-Preserving Hashing for Image Retrieval

Round 1

Reviewer 1 Report

The authors presented a network architecture, which adds a feature enhancement layer to better extract image features, remove redundant features, and express the similarity between images through contrastive loss, thereby constructing compact exact binary code. The paper is well-structured, but the language needs improvements. The introduction will benefit with the presentation of the contributions. The related work must be better explored, and a table with advantages and disadvantages will help to understand the novelty of your work. The proposed method and results are good, but they need to be compared with the literature. What is the novelty presented? Unfortunately, I cannot understand the novelty that must be highlighted in the revision.

Author Response

Thank you very much for your valuable advice.I am very sorry that I did not make you understand the innovation of the paper due to the vagueness of my writing language.Therefore, I add and revise the introduction and related working sections to make my point clear to you.

Author Response File: Author Response.pdf

Reviewer 2 Report

I have the following observations on the article:

1.     The abstract is well written

2.     Some more references latest and relevant to the study need to be included in the Introduction sections

3.     Related work section is quite good where the authors have included the most relevant articles to discuss issues and developments

4.     In section 4.2, the definitions of precision and recall are very famous and common. The author needs to cite some sources for this.

5.     In Table 2, there is a gap between ‘4 8 bit’, correct it.

6.     All figures need to be greatly improved and the labels should be clearly visible.

7.     A more detailed conclusion needs to be written and it must talk about future prospects of your work.

 

I consider this work a significant one and it will definitely add knowledge to the field. Recommended

Author Response

First of all, we are very grateful to you for giving us such specific suggestions for revision. After that, we follow your specific suggestions for revision. We cite the latest references in the additions in the Introduction section, cite sources on the definitions of precision and recall, optimize the labels and numbers that are not clear in the pictures, revise the conclusion section and give directions for future work.

Reviewer 3 Report

This paper presents the approach to learn binary codes using CNN. Generally, the idea is interesting.  However please define more precisely and detailed the additional contribution of the research to the recent state of the research field. The discussion must include the results obtained in comparative analysis. 

Author Response

We are very grateful to you for your great interest in our ideas and providing us with relevant suggestions. In the introduction and related work, we increase the relevant elaboration of our methods, and add relevant literature to reflect the possibility of our work. In addition, we add the advantages and disadvantages of the experiment in the conclusion part.

Round 2

Reviewer 1 Report

The manuscript is improved, and it can be accepted.