Table Structure Recognition Method Based on Lightweight Network and Channel Attention

Round 1

Reviewer 1 Report

This paper is well-organized except some concerns which need to be addressed.

1) In some figures, such as Fig.6, 8, 9 and 10 etc, the texts are not presented clearly. Please check the figure pixel or change them into tables to make them readable.

2) Please check the reference format. Some references are not shown, like Line 52, Page 2.

3) More literature needs to be reviewed.

4) At the end of section 1, only 2 points of main innovation are listed. It is suggested to list at least 3 points.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript is interesting in terms of scope and results. In addition to a review of language and style. The following points should be revisited:

1. The references do not contain the link to go to their section. In addition, they must be in MDPI style format. Finally, some references or tables are not properly linked, for example in line 52, this message appears "[16Error! Reference source not found.]", also in line 104 the same happens [21Error! Reference source not found.].
2. It is necessary to define the meaning of some abbreviations: LGPMA [Line 38], TGRNet [Line 64], RCANet [Line 90], among others.
3. Introduction section should add a last paragraph describing the structure of the document.
4. It is necessary to refer to the Figures, since they are present but not mentioned, which causes the reading thread to be lost.
5. In lines 163 to 170 the entry of an AAR is described, however, it is suggested to rewrite each one of the elements that conform it, with a more detailed wording. For example: "First, the input of the RRA and CAA modules is divided into two parts...; likewise, the number of channels is twice..."
6. The sub-figures within Figure 6 are somewhat difficult to understand, I suggest improving them and adding a little more context at the caption of the figure.
7. The conclusion is too short, I suggest adding how LRCAANe contributes to the state-of-the-art, in relation to other works. Here again, I suggest you add a paragraph with future research directions.

Regards

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

1. Why the authors utilized ResNet18 ,  ShuffleNetv2 model instead of other transfer learning model.

2. Please desribe the tiling operations of the RAA module and the CAA module.

3. Please enhance the resolution of figures for better readability.

4. Discuss more about the findings of your work, why your work is more accurate as compare to the existing work using different parameters.

5. More recent references can be added for more up to date work.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thank you for reviewing the points suggested in the first round. The article can be submitted in its current format.