Scene Recognition Based on Recurrent Memorized Attention Network

Round 1

Reviewer 1 Report

This paper presents a new scene recognition framework, called Recurrent Memorized Attention Network (RMAN). The model performs object-based scene classification by recurrently locating and memorizing objects in the image. The algorithm was tested on two datasets, and the achieved results were compared with the results obtained by other methods known from the literature.

The presented method is interesting, and the paper is well written. I have only a few comments.

Remarks:

1. The authors should carry out thorough proofreading. There are many editorial errors in the paper.
2. Formatting and numbering of equations should be improved. Each equation should have a separate number. According to what pattern is the alignment of equations (7) and (8) made? This should be corrected.
3. The results for the RMAN_fc + SVM method are missing in table 1.
4. Literature is outdated. A few papers from 2019-2020 should be added.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Looking at the title for the first time, I began reading the article with interest. I agree with the authors that presented results proposed a framework for end-to-end scene recognition task and RMAN model can achieve better classification performance (Table 1). Below are suggestions to authors:

• To improve the quality of paper please include a flowchart. 
• I would suggest reducing the number of subsections (3.4.1, 3.4.2) please include it in section 3.4
• please read carefully paper and correct spaces (e.g line 112)
• line 77 - please correct brackets [4] and space between references
• Reference should be corrected according to the guidelines of the Electronics Journal (please read carefully instruction for authors and correct them).
• line 125 - I suggest use [27-29], 128 [30-32]
• line 470 - what kind of criteria, please specify it. 
• It would be great to add dataset for the audition. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript is interesting and the topic is the state of the art of the domain. English is good and it is easy to read. However, I have a concern related to the experiments: the authors make tests with scenarios 30 and 67 comparing with other existing approaches. I do not why these two scenarios were selected. Could the authors show more results of other scenarios? Can they make a mean or similar to validate the quality of the proposal? I think more experiments or one more extensive is necessary. 

 

I hope the authors could solve the lack of experiments and comparisons of the manuscript. The rest is acceptable.

 

I desire the best of lucks to authors in order to improve the quality of the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf