DS-DETR: A Model for Tomato Leaf Disease Segmentation and Damage Evaluation
Round 1
Reviewer 1 Report
This manuscript reports the Transformer-based tomato leaf disease segmentation whose topic, I believe, is relevant to the journal. The challenges of crop diseases are well described, and the overall research design is appropriate. However, several aspects of the manuscript need to be improved. I advise the authors to take into consideration of the following major remarks in order to improve the flow and quality of the presentation of their work. I also suggest to proofread the manuscript carefully as some typos are found and some sentences are difficult to understand.
1. Introduction: the background of the study is well described and supported by relevant and current literature. All relevant information is included, but the objectives of the study are not clearly stand out. I would suggest that the authors attempt to present the key objectives with an emphasised flow, with regards to what are the new findings of the study in terms of scientific knowledge, and thus highlighting the added value of the paper.
Besides, there are too many details about the research methods in the introduction (line 103-142). These details are better fit in the materials and methods.
2. Line 109, 152, 162: Disease Segmentation Detection Transformer (DS-DETR), plant disease classification dataset (PDCD), tomato leaf disease segmentation dataset (TLDSD)
a. Use of abbreviation: some seems to be necessary, but some others seem not. What are the added values of using abbreviation, such as PDCD, TLDSD, where they simply indicate the dataset?
b. In some sentences, each word is capitalized (e.g., line 109), but in some other sentences, no word is capitalized (e.g., line 152 & 162). If authors want to use the abbreviation throughout the manuscript, I suggest having each word capitalized when they are first introduced in the paper.
3. Line 182-185: Figure label needs to be right below the figure.
4. Line 256: The calculation formula is as Eq.2 (not Eq.3). In Equation 2, there is no description what PE stands for.
5. Line 288-289: Ubuntu 18.04, Python 3.9, Pytorch=1.7.0. Pytorch 1.7.0?
6. Line 351-353: there is no equation 6 (Eq.6). Eq. 8 is not about AP, but recall. Equation numbers do not match as were described in the text.
7. Line 152: AI Challenge2018 -? AI Challenge 2018?, Plantvillage -> PlantVillage?
8. Line 397: Table 1. Errata: Tiem(s) -> Time(s)
9. Line 496: Reference 7, 24. Please double check.
Author Response
Dear Editors and Reviewers:
Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “DS-DETR： A Model for Tomato Leaf Disease Segmentation and Damage Evaluation” (Manuscript ID: agronomy-1849465). In the following pages are our point-by-point responses to each of the comments which we hope meet with approval. The main corrections in the paper and responses to the reviewer’s comments are as flowing. The revisions to the manuscript have been marked up using the “Track Changes” function and marked with yellow.
1. Introduction: the background of the study is well described and supported by relevant and current literature. All relevant information is included, but the objectives of the study are not clearly stand out. I would suggest that the authors attempt to present the key objectives with an emphasized flow, with regards to what are the new findings of the study in terms of scientific knowledge, and thus highlighting the added value of the paper.
Besides, there are too many details about the research methods in the introduction (line 103-142). These details are better fit in the materials and methods.
Response:
â‘ According to reviewers' comments, we have modified part of the introduction to emphasize the study of leaves and leaf disease segmentation. And we have rewritten the last paragraph of the introduction to re-emphasize the objectives and main contributions to the study of leaf disease segmentation. See Line 81-84, 98, and 103-125 for details.
â‘¡ According to reviewers' comments, we have rewritten Line 103-142 to simplify the part of research methods in the introduction. And we have put more details into the materials and methods.
2. Line 109, 152, 162: Disease Segmentation Detection Transformer (DS-DETR), plant disease classification dataset (PDCD), tomato leaf disease segmentation dataset (TLDSD)
a.Use of abbreviation: some seems to be necessary, but some others seem not. What are the added values of using abbreviation, such as PDCD, TLDSD, where they simply indicate the dataset?
b.In some sentences, each word is capitalized (e.g., line 109), but in some other sentences, no word is capitalized (e.g., line 152 & 162). If authors want to use the abbreviation throughout the manuscript, I suggest having each word capitalized when they are first introduced in the paper.
Response:
â‘ PDCD and TLDSD represent two different datasets, which are applied to different tasks. We abbreviated the two datasets for simplicity and readability. Thank you for your suggestions. We realize that too-long abbreviations can also cause reading difficulties. Therefore, we changed the abbreviation of TLDSD to TDSD to indicate the target and application.
â‘¡ Many thanks for your suggestions. We reviewed and revised the abbreviations of the paper. We capitalize each word when we first introduce it in the paper.We have modified Line 19, 22, 29 54, 59, 64, 103, 115, 117, 135, 145, 146, 163, 173, 186, 221, 324-329 of the paper.
3.Line 182-185: Figure label needs to be right below the figure.
Response:
Many thanks for pointing out the error. We have modified Figure 2 and the description of Figure 2.
4. Line 256: The calculation formula is as Eq.2 (not Eq.3). In Equation 2, there is no description what PE stands for.
Response:
â‘ Many thanks for pointing out the error. We have modified Eq.2 in Line 235.
â‘¡ PE represents position encoding in Equation 2. We have added the description of PE in Line 236.
5. Line 288-289: Ubuntu 18.04, Python 3.9, Pytorch=1.7.0. Pytorch 1.7.0?
Response：
We revisited the experimental environment. Our experimental environment is Python=3.7, Pytorch=1.7.0. Our original code reference https://github.com/facebookresearch/detr/tree/master. Run the link code, requiring Pytorch >= 1.5.0. We have modified Line 266.
6. Line 351-353: there is no equation 6 (Eq.6). Eq. 8 is not about AP, but recall. Equation numbers do not match as were described in the text.
Response：
Many thanks for pointing out the error. We revisited the Equation numbers. Eq. 6 and Eq. 7 are about precision and recall. And Eq. 8 is a description of AP. We have modified the Equation numbers 7, 8, 9 to 6, 7, 8.
7. Line 152: AI Challenge2018 -? AI Challenge 2018?, Plantvillage -> PlantVillage?
Response：
Many thanks for pointing out the error. We have modified AI Challenge2018 to AI Challenge 2018 and Plantvillage to PlantVillage in Line 135.
8. Line 397: Table 1. Errata: Tiem(s) -> Time(s)
Response：
Many thanks for pointing out the error. We have modified Tiem(s) to Time(s) in Table 1.
9. Line 496: Reference 7, 24. Please double check.
Response：
Many thanks for your suggestion. We have modified References 7, and 24 in Line 479 and 513.
Author Response File: 
Author Response.doc
Reviewer 2 Report
This paper has an interesting topic but due to a lack of structure, I suggest a major review.
In this paper, the structure is in appropriate form. The Result section is awkwardly written. There is no need to write the same values both in tables and in the text. Moreover, there is no Discussion section in this paper. The Conclusion is very briefly written and future improvements are not stated.
Please remove the link to the GitHub repository from the Abstract.
In Figures, there is no need for writing both labels below the image and in the description where it has to be.
Description of Figure 2 is not below the Figure.
Specific comments:
line 113-114: Please remove this sentence: „Specifically, I have done the following work.”
line 256: It should be written Eq.2.
line 358-359: Please explain what higher AP means with professional terms.
Author Response
Dear Editors and Reviewers:
Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “DS-DETR： A Model for Tomato Leaf Disease Segmentation and Damage Evaluation” (Manuscript ID: agronomy-1849465). In the following pages are our point-by-point responses to each of the comments which we hope meet with approval. The main corrections in the paper and responses to the reviewer’s comments are as flowing. The revisions to the manuscript have been marked up using the “Track Changes” function and marked with yellow.
1. This paper has an interesting topic but due to a lack of structure, I suggest a major review. In this paper, the structure is in appropriate form. The Result section is awkwardly written. There is no need to write the same values both in tables and in the text. Moreover, there is no Discussion section in this paper. The Conclusion is very briefly written and future improvements are not stated.
Response:
Thanks for your suggestions. According to the Reviewer’s suggestion, we have reviewed our structure. We have modified our Results and Conclusion. And we have added Discussion section to this paper.
â‘ We are sorry that our results and tables have repeatedly expressed specific values. Thank the reviewers for their comments. We have removed some duplicate values and supplemented the data analysis in the results. We have modified some statements in Line 350-352, 353-355, 364-365, 381-383, and 392-394.
â‘¡We have added Discussion section in this paper. In Conclusion and Discussion, we have recapitulated the research contents and contributions of this paper. The existing problems of the model are discussed and future work is expected. We have modified some statements in Line 427-450.
2. Please remove the link to the GitHub repository from the Abstract.
Response:
According to the Reviewer’s suggestion, we have removed the link to the GitHub repository from the Abstract in Line 33-34.
3. line 113-114: Please remove this sentence: “Specifically, I have done the following work.”.
Response:
According to the Reviewer’s suggestion, we have removed this sentence: “Specifically, I have done the following work.” in Line 113-114.
4. In Figures, there is no need for writing both labels below the image and in the description where it has to be.
Response:
According to the Reviewer’s suggestion, we have removed labels in the description in Figure 1 and Figure 7.
5. Description of Figure 2 is not below the Figure.
Response:
Many thanks for pointing out the error. We have modified Figure 2 and the description of Figure 2.
6. line 256: It should be written Eq.2.
Response：
Many thanks for pointing out the error. We have modified Eq.2 in Line 235.
7. line 358-359: Please explain what higher AP means with professional terms.
Response：
We may not have a clear explanation of what higher AP means. The Average Precision (AP) is the area under the precision-recall (P-R) curve obtained for the number of correctly predicted pixels to the number of pixels in the ground truth for a single class. The higher AP refers to a higher AP value. The higher AP value, the more correct targets are predicted. We have modified some statements in Line 328, 333-335,338-339 to better describe the AP value. Thanks for your questions.
Author Response File: Author Response.doc
