MediaPipe’s Landmarks with RNN for Dynamic Sign Language Recognition
, Abanoub R. Wadie 1, Abanoub K. Attia 1, Abanoub M. Asaad 1, Andrew E. Kamel 1, Salwa O. Slim 1, Mohamed S. Abdallah 2,3,* and Young-Im Cho 2,*
Round 1
Reviewer 1 Report
The key contribution of this work should be conveniently explained.
MediaPipe is presented as the grounding work but it should not be supported on a non-published preprint.
RNN and DSL acronyms should be presented.
The sentence in line 9 is not correct: "DSL10-Dataset was created in response to the scarcity of English video-based datasets."
Models in line 61 require references.
The first sentence in line 72 should be fixed.
The dataset in line 140 should be fixed.
Figures 1 a b and c. are in low resolution and stretched.
Figure 2 is stretched and requires a reference to the original work. Please, check the other ones.
Figure 3 requires a reference.
The sentence in line 169 "The MediaPipe framework provides a solution for most of the DSL problems."
DSL10-Dataset requires a reference.
The link at line 273 should be replaced by a reference.
The size of the font in the Tables is too huge.
Figure 8 includes stretched images.
Computations on lines 163 and 165 should be replaced by formulas.
This statement requires further discussion "This experiment shows the pros and cons of including face keypoints.".
The dataset is not correctly described and does not include labels.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Author,
I have some comments on your article:
1. In section: 3.1. Input Data - please write more about the input data and how to obtain them
2. Please improve the quality of figure 1 to figure 3
3. Literature should be checked if there are no newer items. Especially from the last 18 months. It would be good to add several references.
4. Please correct the text formatting in the tables
5. Experimental Results - Please describe the results in more detail
6. In the summary of the article, please provide more technical information about the practical possibility of implementing the proposed method.
Best regards
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper has been improved with the and all previous comments have been addressed.
Reviewer 2 Report
Dear Authors,
Thank you very much for introducing changes that have improved the quality of the article. I have no more comments.
Best regards
