MRS-Transformer: Texture Splicing Method to Remove Defects in Solid Wood Board

Round 1

Reviewer 1 Report

Please elaborate on why a two-group RSwin model was used with four layers of depth for the decoder.

Additionally, it would be informative to include a breakdown of the computational cost, especially since the MRS-Transformer algorithm was able to complete a single wood texture image five times faster than other algorithms.

I kindly request more details to be included in Table 2.

Based on the experimental results that demonstrate the superiority of the MRS-Transformer algorithm over DeepFill v2 and TG-Net algorithms in terms of both objective metrics and visual outcomes, it would be beneficial to provide more detailed information.

Author Response

Dear reviewer,

Many thanks for your constructive remarks. We adapted our manuscript, to answer your concerns. All changes are marked in track changes in the document attached. Note that we received feedback from other reviewers as well, which is also reflected in these changes in the manuscript.

Best wishes,

The authors

 

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper focuses on defect removal and image inpainting in solid wood boards. For this, an asymmetric encoder-decoder architecture based on Vision Transformer is presented.  A reverse Swin (RS) module is designed to adjust the size of divided image patches and complete the missing patches.  The authors compared the proposed work with DeepFill v2 and TG-Net algorithms. The overall structure of the paper is good. However, there are issues that must be addressed.

There are some concepts that are briefly mentioned but are not explained at all, especially the concept of NLP tasks. Information about NLP tasks should be provided.

No information is provided regarding implementation, which platform is used, etc.

This paper does not discuss the limitation of the proposed work. A separate section is required for this.

The paper has several grammatical issues that should be corrected. For instance

"splicing ." ->  splicing.

 

"Chen et al.[13]proposed" -> Chen et al. [13] proposed

Some abbreviations and terms should be explained, such as weighted L2 loss, MLP

 

Author Response

Dear reviewer,

Many thanks for your constructive remarks. We adapted our manuscript, to answer your concerns. All changes are marked in track changes in the document attached. Note that we received feedback from other reviewers as well, which is also reflected in these changes in the manuscript.

Best wishes,

The authors

Author Response File: Author Response.pdf