A Real-Time Nut-Type Classifier Application Using Transfer Learning
Round 1
Reviewer 1 Report
The author has tried to apply AI for the nuts clasification using Transfer Learning (TL) techniques. In the process, four open source deep learning classifiers namely ResNet, EfficientNet, Inception, MobileNet and a custom CNN model have been used. The classifiers' performance evalaution is based on parameters- validation loss, validation accuracy and F1-score. The selection of TL classifiers is understandable and reasonable. But, the application for which the whole exercise is done is not convincingly justified.
Nowadays, the nuts are already avaliable in a packed form. What is the necessity of classifying the nuts? If it is really needed or it is a demand from the merchants/ traders, then it should be properly conveyed as a research problem. Applyying technology or the softwares is good, but applying them in an application in which they are not needed is a waste of time and efforts.
Although, the TL methods and classifiers address the use of AI but the application (nuts classification) for which they were used is inappropriate.
There are several instances in which English language corrections are needed. For example, in lines 14-15, "we propose to use ResNet, EfficientNet, Inception, MobileNet architectures...". It should be framed as " Architectures like ResNet, EfficientNet, Inception, MobileNet have been used..."
In lines 371-372, "Hence, the effect of the number of labels in the dataset can be further examined even this study investigates the performance of a multi-classification solution." The sentence re-framing is required.
Author Response
The response to the reviewer is attached as a file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Accept the article in its present form.
Author Response
The response to the reviewer is attached as a file.
Author Response File: Author Response.pdf
Reviewer 3 Report
Overall the paper is well written despite the very limited scientific significance. Reference should be provided for equations 1-4.
Author Response
The response to the reviewer is attached as a file.
Author Response File: Author Response.pdf
Reviewer 4 Report
· The introduction needs to be rewritten to make it more specific and engaging. It should also include a justification for the need for nut classification models and their applications, as well as a discussion of the research gaps in previous studies in this area.
· Some of the sections in Section 3 should be moved to the previous section, such as the Performance Evaluation section.
· The explanations of the methods and techniques used in the study are adequate.
· The section heading for Section 3 should be changed to "Results and Discussion". The Discussion section should be removed.
· The paper needs to be proofread carefully, as some parts are difficult to understand.
The paper needs to be proofread carefully, as some parts are difficult to understand.
Author Response
The response to the reviewer is attached as a file.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have incorporated the suggestions given by the reviewers; hence, the manuscript can be accepted.
Author Response
The response to the reviewer is attached as a file.
Author Response File: 
Author Response.docx
Reviewer 4 Report
Fig 4, 5, 6, & 7 resolution is too low
This comment not implemented " Some of the sections in Section 3 should be moved to the previous section, such as the Performance Evaluation section. " and the revised structure would look like.
Introduction
2. Materials and Methods
3. Results and Discussion
4. Conclusions
The paper needs to be proofread carefully, as some parts are difficult to understand in the revision.
Author Response
The response to the reviewer is attached as a file.
Author Response File: Author Response.pdf
