Classification of Grain Storage Inventory Modes Based on Temperature Contour Map of Grain Bulk Using Back Propagation Neural Network
Round 1
Reviewer 1 Report
In this paper, the authors propose a method of using a temperature contour map converted from digital temperature data to classify stored grain inventory modes in a large bulk grain warehouse, which mainly included detect inventory changes, and routine operations performed.
The article has a good flow and readability. Moreover, it has a good perspective but needs many improvements and clarifications.
First of all, the text has good use of the English language but, some syntax errors need correction.
I suggest the authors revise the title of the paper. Please choose a title that reflects the work conducted in this paper, be clear and informative.
The abstract and introduction are appropriately written. The introduction provides sufficient background with up-to-date literature.
Please add a separate section entitled "Related Work" to transit smoothly in the following parts of the work. Also, evaluate how your study is different from others compared with other state-of-the-art approaches.
The Methodology section includes an extensive and detailed description of the preferred methods.
I'd like the figures to have better alignment(in the middle).
Figures 5, 7 and 8 contain a lot of information. It should be included in the body of the text.
The authors miss the experiment setup. Please, demonstrate the environment of the experiment in detail.
Try to make a more coherent, accurate and focused presentation of the results.
For discussion, I would like to know the limitations and the potential issues of this study.
The conclusion presented in this study should be more advanced and, the authors should report the advantages of the proposed approach in more detail. Moreover, conclusions can discuss future research directions and extensions of the study.
I recommend the authors include more recent references from MDPI.
To sum up, the authors should further highlight the preferred methods as the selected approach did not allow yet to illustrate the value of the proposed model in the experimental research.
Author Response
Reviewer #1.: Manuscript - agriculture-1178772
Thank you for your comments. We have revised the manuscript according to your comments.
Comment 1
First of all, the text has good use of the English language but, some syntax errors need correction.
Response
The text has been checked and some syntax errors has been corrected.
Comment 2
I suggest the authors revise the title of the paper. Please choose a title that reflects the work conducted in this paper, be clear and informative. The abstract and introduction are appropriately written. The introduction provides sufficient background with up-to-date literature.
Response
After thinking twice, we think the title ‘Classification of grain storage inventory modes based on temperature contour map of grain bulk using back propagation neural network’ can reflect the work conducted in this paper.
Comment 3
Please add a separate section entitled "Related Work" to transit smoothly in the following parts of the work. Also, evaluate how your study is different from others compared with other state-of-the-art approaches.
Response
‘Related Work’ had been written in introduction. For reader better to comprehend, we rearrange the contents as a separate section Line 41-50, Also, the difference has been added Line 47-50.
Comment 4
The Methodology section includes an extensive and detailed description of the preferred methods.
Response
The BP neural network is used most widely and successfully for classification. Many literatures have introduced the principle of BP neural network, so we don't think it is necessary to add additional and detailed description of the BP neural network here. We have added relevant references. Line 222.
Comment 5
I'd like the figures to have better alignment(in the middle).
Response
All figures alignment has been checked and corrected.
Comment 6
Figures 5, 7 and 8 contain a lot of information. It should be included in the body of the text.
Response
The more information has been added. Line 296-305, Line 345-353, Line 376-396.
Comment 7
The authors miss the experiment setup. Please, demonstrate the environment of the experiment in detail.
Response
The hardware environment of the models the experiment has been added in Line 244-246. Too many locations of grain warehouse, the detail about the environment of these warehouse is too many to describe in detail. Thus, the general environment of these granaries is introduced in Line 156-159.
Comment 8
Try to make a more coherent, accurate and focused presentation of the results.
Response
The results in section 3 have been re-written.
Comment 9
For discussion, I would like to know the limitations and the potential issues of this study.
Response
The limitations and the potential issues of this study has been added in Line 436-444.
Comment 10
The conclusion presented in this study should be more advanced and, the authors should report the advantages of the proposed approach in more detail. Moreover, conclusions can discuss future research directions and extensions of the study.
Response
The conclusion has been re-written Line 445-463.
Comment 11
I recommend the authors include more recent references from MDPI.
Response
More recent references from MDPI has been added. Reference 5, 6, 15, 18, 21, 22, 25, 26, 39, 44, 53.
Reviewer 2 Report
The authors propose a machine-learning based method to quantify grain conditions based on digital temperature data inside of grain depots. Overall, the study is well designed and the methods are well executed. I would suggest that the authors consider adding an additional model or two, in order to demonstrate the accuracy of the neural network model. In particular, it would be useful to include a null model that uses the average recorded digital temperature data in order to classify grain inventory modes. Apart from a null model, it would be useful to include a model with strengths that are complimentary to a neural network. E.g., using generalized additive models that offer much better explanability compared to neural networks. Secondly, it would be useful to provide some background on why the specific features were used (color coherence vectors etc.). This would be helpful for the reader to figure out if some other features could perform better.Author Response
Reviewer #2 : Manuscript - agriculture-1178772
Thank you for your comments. We have revised the manuscript according to your comments.
Comment 1
I would suggest that the authors consider adding an additional model or two, in order to demonstrate the accuracy of the neural network model. In particular, it would be useful to include a null model that uses the average recorded digital temperature data in order to classify grain inventory modes. Apart from a null model, it would be useful to include a model with strengths that are complimentary to a neural network. E.g., using generalized additive models that offer much better explanability compared to neural networks.
Response
This comment is great. Since this article tends to be applied on site, thus we added a commonly used classification method - SVM as the additional model to compared to neural networks. Line 276-283, Line 424-435.
Comment 2
Secondly, it would be useful to provide some background on why the specific features were used (color coherence vectors etc.). This would be helpful for the reader to figure out if some other features could perform better.
Response
We have added some background on why the specific features were used. Line 176-182, 196-200, 210-212.
Reviewer 3 Report
Agriculture (ISSN 2077-0472)
Manuscript Number: agriculture-1178772
Title: Classification of grain storage inventory conditions based on temperature contour map of grain bulk using back propagation neural network
Article Type: Article
The subject of research includes in this journal. Research work is interesting. In the paper titled “Classification of grain storage inventory conditions based on temperature contour map of grain bulk using back propagation neural network”.
The authors used one of the artificial intelligence methods which are artificial neural networks for the classification. The approach is novel and interesting. The authors have developed a method whose classification efficiency is quite good. The paper needs minor revisions and need to take care of following:
- The quality of the language is insufficient. Have a native speaker or similar assist you.
- The introduction requires a larger review of the world literature. Because in scientific databases you can find similar scientific articles that are not cited.
- The authors did not state what number of inputs were used for the study.
- Please provide the exact types of temperature measurement sensors?
- Is empty bin, was the combination (SFV) implemented as it is not visible in figure 4
- Why did the authors not take grain humidity? Does this parameter have no influence?
- The conclusions should be rewritten and more extended.
In my recommendation is minor revision.
Author Response
Reviewer # 3:Manuscript - agriculture-1178772
Thank you for your comments. We have revised the manuscript according to your comments.
Comment 1
The quality of the language is insufficient. Have a native speaker or similar assist you.
Response: The language has been polished by one author who is professor in the University of Manitoba, Canada.
Comment 2
The introduction requires a larger review of the world literature. Because in scientific databases you can find similar scientific articles that are not cited.
Response
The introduction has been added more and larger review of the world literature.
Comment 3
The authors did not state what number of inputs were used for the study.
Response
The number of inputs samples was the 85% of the total dataset shown in Line 168.
Comment 4
Please provide the exact types of temperature measurement sensors?
Response
Different grain storage warehouse used different types of temperature measurement sensors from different manufacturer. Then, we can not provide the exact types of temperature measurement sensors. However the sensors had accuracies typically within 1℃, as per the standard of State Grain Administration. This is clarified with an added reference (Line 153-155).
Comment 5
Is empty bin, was the combination (SFV) implemented as it is not visible in figure 4
Response
It is not empty bin. The bin N in the CCV is 60, 70, 80 (Line 257). Figure 4 has been revised. Six matrices were implemented as input matrix of neural network individually, including CCV, TFV, SFV, combination matrix of CCV and TFV (CCV&TFV), combination matrix of CCV and SFV (CCV&SFV), combination matrix of CCV and TFV and SFV (CCV&TFV&SFV) (Line 253-255).
Comment 6
Why did the authors not take grain humidity? Does this parameter have no influence?
Response
This research attempted to use temperature data alone to assess the grain conditions. Furthermore, the data sets were from the actual grain depots and grain humidity data was not collected in these facilities. This parameter have no influence on the detection results.
Comment 7
The conclusions should be rewritten and more extended.
Response
The conclusions has been rewritten and extended. Line444-462.
Round 2
Reviewer 1 Report
