Novel Assessment of Region-Based CNNs for Detecting Monocot/Dicot Weeds in Dense Field Environments
Round 1
Reviewer 1 Report
Overall the work the good done by the authors are good. There are few things that need to be improved.
1) Description of Fig 2, 5 and 6 to be improved with more details.
2) What is PPPDD and PPPMM in Section 4.3.1?
3) Datasets used in this paper are publically available? if not are the author willing to share the data for the research community to reproduce the results
Author Response
Dear Reviewer,
Thank you for your valuable comments and suggestions,
A word file "ResponseToReviewers" has been uploaded. All the questions and the comments have been responded in this file.
Best regards,
Nima Teimouri
Author Response File: Author Response.pdf
Reviewer 2 Report
The authors presented results of weed recognition using neural networks. The article was prepared correctly, the authors drew the right conclusions.
The study lacks a discussion of the results - a comparison of the obtained results with other authors.
Author Response
Dear Reviewer,
Thank you for your valuable comments and suggestions,
A word file "ResponseToReviewers" has been uploaded. All the questions and the comments have been responded in this file.
Best regards,
Nima Teimouri
Author Response File: Author Response.pdf
Reviewer 3 Report
please see attached file.
Comments for author File: Comments.pdf
Author Response
Dear Reviewer,
Thank you for your valuable comments and suggestions,
A word file "ResponseToReviewers" has been uploaded. All the questions and the comments have been responded in this file.
Best regards,
Nima Teimouri
Author Response File: Author Response.pdf
Reviewer 4 Report
This study presents a CNN-based approaches to detect monocot/dicot weeds in fields. Both EfficientDet and YOLOv5 are used for the detection task. A novel image processing method is developed to generate synthetic images test the performance of the model. This study also provide a dataset publicly for future researches. The paper is well prepared and has some new contributions, especially the use of synthetic data to consider realistic situation. In fact, the use of synthetic data generation could also covercome the limitation of small number dataset and to avoid overfitting in realistic applications. I have some comments, as follows:
- There are also other state-of-the-art networks such as Faster RCNN? How about a comparision with this model
- How about the effect of image resolution on the performance of the CNN-base detectors?
- Figure 12, it is hard to distinguish the detected objects.
Author Response
Dear Reviewer,
Thank you for your valuable comments and suggestions,
A word file "ResponseToReviewers" has been uploaded. All the questions and the comments have been responded in this file.
Best regards,
Nima Teimouri
Author Response File: Author Response.pdf