Next Article in Journal
CSGN: Combined Channel- and Spatial-Wise Dynamic Gating Architecture for Convolutional Neural Networks
Previous Article in Journal
Machine Learning-Enabled Internet of Things (IoT): Data, Applications, and Industry Perspective
 
 
Article
Peer-Review Record

Grape Maturity Detection and Visual Pre-Positioning Based on Improved YOLOv4

Electronics 2022, 11(17), 2677; https://doi.org/10.3390/electronics11172677
by Chang Qiu 1, Guangzhao Tian 2,*, Jiawei Zhao 2, Qin Liu 3, Shangjie Xie 2 and Kui Zheng 4
Reviewer 1:
Reviewer 2:
Electronics 2022, 11(17), 2677; https://doi.org/10.3390/electronics11172677
Submission received: 27 July 2022 / Revised: 23 August 2022 / Accepted: 23 August 2022 / Published: 26 August 2022

Round 1

Reviewer 1 Report

The introduction, methods, experiments, and results were well presented.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper is calledGrape maturity detection and visual pre-positioning based on improved YOLOv4“.

 

Some specific comments are as follows:

 

- The literature on the subject should be increased in the study. In addition, the results should be discussed with the support of the literature.

-The use of a single variety may be sufficient for current conditions. But there are too many grape varieties in the world. Therefore, the existence of grape varieties with different characteristics is very important in terms of the health of the study and its access to large areas.

 

-The work is planned for harvesting. For this, algorithms were used according to 4 different maturity levels. As it is known, grapes reach harvest maturity at the same time. In this case, 2 periods can be used as harvest or close to harvest. So you can work with more images. In addition, the success of the algorithm may increase. You can even work with different types. Thus, it is possible to obtain realistic results.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript presents an improved version of the YOLOv4 model for grape maturity detection (SM-YOLOv4). In addition, a pre-positioning method is also proposed. The paper is well-written, and the authors clearly explain the proposed techniques. However, I think the authors should include a background section in which the models used for comparison (YOLOv5, SM-YOLOv4, YOLOv4-Tiny, and Faster_ R-CNN) are briefly described.

The results of the improved YOLOv4 have been compared with other techniques, including the more recent YOLOv5. I have some doubts about this point:

1) Is it possible to apply the idea of the proposed improvement to YOLOv5? If so, why have the authors chosen YOLOv4?

2) In the comparison of the training results (see Table 2), the authors have not included YOLOv4 (like in Table 3); what is the reason?

Regarding the pre-positioned results, I think that the relationship between the improvements obtained by SM-YOLO4 and those obtained by the pre-positioning method should be explained.

There are some minor mistakes, e.g., the unit of speed in Tables 2 and 3 (I think it is an ideogram that has not been translated).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Back to TopTop