Weed Detection in Potato Fields Based on Improved YOLOv4: Optimal Speed and Accuracy of Weed Detection in Potato Fields

Round 1

Reviewer 1 Report

1- Sharing the Dataset as open access will be convenient for readers to test.

2- In the study, the hold-out CV method was preferred. What are the results for k-fold CV? For example, for k=10

3- What is the distribution of the dataset by classes? Maybe a table can be shared.

4- The motivation and novelties of the study can be given in a sub-title.

5- Avoid block citations. E.g. 1-4, 5-9, 10-15 etc.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

MDPI Electronics

The following is an overview of the article Weed Detection in Potato Field Based on Improved YOLOv4 (electronics-1993974). In this paper presents a in order to identify various types of weeds in potato fields more accurately and quickly, and to meet the demands of precision agriculture, this study improves on the YOLOv4 algorithm and proposes the MC-YOLOv4 target detection algorithm. The manuscript has contributions to the area of weed identification, YOLOv4, and precision agriculture.

However, some points must be highlighted so that the author(s) can review and submit in another round of review: The following corrections are considered to be beneficial for the strengthening of the article.

1. The Conclusions should be reviewed again. The original aspect of the study and its difference from other studies should be clearly explained. (The conclusion should be explored better and it needs to contemplate the eventual restrictions of the developed technique to address future works in this area.)

2. The abstract must be make strong. Abstract should be reviewed again.

3. Some sentences have spelling errors. (Punctuation marks, spaces, etc.). Some places should be left space. Please check all sentences in article.

4. It has been a comprehensive study in the literature in recent years. If there are more current literature studies, these should be examined in detail and added to the literature section.

5. The authors should compare the results of their method with those of previous studies. As mentioned in the literature, there are several methods with very high accuracy, even better than the proposed method. Author(s) can do compare table (A new table can add about previous studies to result section.). This subject is very important.

6. The motivations of the proposed method are not clear. Which problem does the proposed method attempt to solve? Why the other existing diagnosis methods failed to solve it? What are the advantages of the proposed method compared to other methods? Those should be illustrated more clearly.

7. Carefully check all grammatical error. Still, the English language should be improved. I suggest asking for help from a native English.

 

I think it is ACCEPTABLE after the MAJOR Revisions mentioned.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors tackle in their paper the weed detection in Potato field. The proposed method try to deal with identification of weed In corps, this method based on YOLOv4 algorithm. The main idea to solve the problem with the authors fused Convolutional Block Attention Module into the PANet structure. The authors proposed MC-YOLOv4 to improve the classic version. The proposed model gives a promising result.

The authors presented an overview for the state of the art.

The contributions of the research are presented.

The paper in current state can be considered for publication, yet there are some changes need to be carried out for final version.

·         Abstract section is too long, the authors should give brief abstract which is usually between 150 and 250 words.

·         Some parts of the paper lack of references.

·         the authors should give an explanation of variables used in each equation.

·         The authors should use linking in paragraphs such as figures, tables and references, to make an easy access for each resource.

·         The author should avoid using two consecutive titles, it is better to give a small introduction after each title.

·         Figure 14 shows some results with YOLOv4 however the values presented in each one of them unreadable.

·         The authors show the loss function graph for their proposed model, yet there is not the one presents the accuracy.

·          The authors in their paper gives some metrics used to evaluate the model, yet there are no tables shows the final values.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Detection of weed using YOLOv4 is presented. The model is improved by replacing DarkNet53 with MobileNetv3 network and K-means++ clustering algorithm. I have some major and minor comments that need to be addressed. 

Major comments:

1. Similar research has been already published with similar approach. The only difference is Soft-NMS for K-means++ clustering algorithm. What is new technical contribution in terms of methodology to the published article,

''Pei, H., Sun, Y., Huang, H., Zhang, W., Sheng, J. and Zhang, Z., 2022. Weed Detection in Maize Fields by UAV Images Based on Crop Row Preprocessing and Improved YOLOv4. Agriculture, 12(7), p.975.''

This need to be checked as similar research work is already there.

2. Replacing Mobilenetv3 and employing CBAM cannot be considered as major contribution. In the paper, it should be clear that it is application wise novel or system model wise novel. 

Minor comments:

1. The abstract need to be checked as it should be concise and precise highlighting the technical contribution or application wise contribution of the paper. Mistake such as CIOU need to be checked and abbreviated similar to CBAM. 

2. In the introduction section, the application of deep learning needs to be cited from different perspectives and the following paper need to be cited to cover the study, ''Hassan, S.A., Rahim, T. and Shin, S.Y., 2021. An improved deep convolutional neural network-based autonomous road inspection scheme using unmanned aerial vehicles. Electronics, 10(22), p.2764''. 

''Roy, S., Menapace, W., Oei, S., Luijten, B., Fini, E., Saltori, C., Huijben, I., Chennakeshava, N., Mento, F., Sentelli, A. and Peschiera, E., 2020. Deep learning for classification and localization of COVID-19 markers in point-of-care lung ultrasound. IEEE transactions on medical imaging, 39(8), pp.2676-2687''.

3. Provide the parameter in a table for configuring the proposed model. 

4. The need of using H-SWISH need to be defined. What advantage it provides using over MISH etc. 

5. In the paper, deep learning, convolutional neural network etc. have been used. It's better to abbreviate it as DL and CNN and use in all paper. 

6. Missing references for losses such GIoU, IoU, etc. The papers need to be cited. 

7. Eq (2), it is IoU not IOU. Also, in eq (3) what is alpha, what it stands for and act as in equation. 

8. Typos error in eq (6), learning rate is represented by eta. check this. What is p in this equation. 

9. Sentences such as, The bneck structure also introduces a light-weight attention mechanism model.......................two layers of network. Revise it. 

10. What is beta in eq (7).

11. Lengthy sentences need to be avoided such as in subsection 2.2.6. Also in this subsection, CAM and SAM needs to be abbreviated first. Check this carefully. 

12. Figure 12, GAM or GMP? Check such mistakes all in the paper. 

13. Research work based on YOLO such as the one reported in ''Han, S.H., Rahim, T. and Shin, S.Y., 2021. Detection of faults in solar panels using deep learning. In 2021 International Conference on Electronics, Information, and Communication (ICEIC) (pp. 1-4). IEEE''., considered a light-weight DL model where different conv filter size in each conv layer are employed. Did you consider such scenario rather than replacing backbone and NMS?

The paper needs to substantial revision in terms of expressional and grammatical mistakes as highlighted above.  

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Electronics (ISSN 2079-9292) | An Open Access Journal from MDPI

 

Dear Editor;

The author(s) made all the corrections mentioned (electronics-1993974 - Weed Detection in Potato Field Based on Improved YOLOv4).

The length of the paper is enough in terms of a scientific paper. Considering studies conducted and results obtained, it is believed that the paper is eligible to be published in your journal after your approval.

 

I think it is ACCEPTABLE  in your journal after your approval as editor.

 

 

 

Comments for author File: Comments.pdf

Reviewer 4 Report

The authors have addressed most of my concerns and I have no further comments.