Radish Growth Stage Recognition Based on GAN and Deep Transfer Learning
Round 1
Reviewer 1 Report
The article is interesting, but some concerns are
1- The abstract is incomplete and does not include all the necessary parts.
2- A Related Works section should be added, and the Introduction section should be rewritten.
3- Article needs to add a discussion section.
4- What are the contributions of authors to this article? They must be explained.
5- Few sections need english corrections in the content.
6- The highlighted article deals similar area of research focus, author might consider for literature study. https://doi.org/10.1007/s11042-023-15363-4
The article needs extensive english revision.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Contributions:
1. The paper proposes a model for radish growth stage identification using machine vision, focusing on the recognition of germination, seedling, and leaf vigorous growth stages.
2. The authors compare and test various transfer learning models, including Inception-v3 and MobileNet, to evaluate their performance in terms of accuracy, loss, precision, recall, and F1 score.
3. The authors propose three improved deep transfer learning methods based on the Inception-v3 model, aiming to enhance recognition performance and reduce overfitting.
4. The effectiveness and robustness of the proposed models are verified using collected datasets, including a cherry radish dataset, the Oxford Flower dataset, and a pest and disease classification dataset.
5. The experimental results demonstrate that the proposed models, particularly the improved model 2, outperform the original model in terms of accuracy, loss, and stability.
6. The paper highlights the potential of machine vision in intelligent agriculture and suggests future work to improve the image acquisition system, expand the dataset, and integrate target detection with growth stage identification.
Comments for Improvement:
i. Provide more detailed information about the collected datasets, such as the number of images, image resolution, and any data augmentation techniques used. This information will help readers understand the dataset's characteristics and potential limitations.
ii. Clearly explain the rationale behind selecting the specific transfer learning models (Inception-v3 and MobileNet) for comparison. Discuss their architectural differences, advantages, and suitability for the task at hand. This will strengthen the justification for choosing these models over others.
iii. Describe the training procedure in more detail, including the optimization algorithm used, the number of training epochs, and any hyperparameter settings. This information will help readers reproduce the experiments and understand the training process thoroughly.
iv. Provide a more comprehensive analysis of the experimental results, including statistical significance tests or confidence intervals to support the claim that the proposed models outperform the original model. This will strengthen the credibility of the results and conclusions.
v. Discuss the limitations and potential biases introduced by using a small dataset. Address the potential impact of limited generalization and the need for further validation on larger and more diverse datasets to ensure the robustness of the proposed models.
vi. Expand the discussion on the practical implications and real-world applications of the proposed models. How can the identified growth stages of radish contribute to improved agricultural practices or decision-making processes? This will provide a broader perspective on the significance of the research.
vii. The bibliography must be improved by citing some up-to-date and relevant references. A few relevant references are provided as examples below:
a. https://www.frontiersin.org/articles/10.3389/fpubh.2019.00340/full
b. https://www.mdpi.com/1424-8220/20/21/6174
As a whole, the paper makes valuable contributions by proposing a model for radish growth stage identification, evaluating different transfer learning models, and presenting improved methods. However, addressing the suggested improvements will enhance the clarity, reproducibility, and impact of the research.
Please check the entire paper as there are some grammatical glitches.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The study proposed automatic recognition of cherry radishes. For this purpose, a dataset has been made for radish growth.The growth stage recognition method based on DCGAN for image preprocessing and data enhancement has been proposed.
The main contributions of the paper and add to the subject area compared with other published material :
A deep transfer learning method has been introduced to improve the accuracy of image classification with a small dataset.
The effectiveness of the proposed model has been proved by the dataset collected in the paper, the official dataset of Oxford Flowers, and the pest and disease classification dataset of AI challenger in 2018.
The specific improvements are to apply the deep transfer learning method to the automatic recognition of cherry radishes,to create a dataset, and to give a performance analysis.
The main addressed problem in the paper is to automatic recognition of cherry radishes by the deep transfer learning model. The topic covered is up-to-date. The paper is well-written and well-organized.
Also, the paper has a few grammatical and punctuation errors. "Nomalization, F1Score,...", especially in Figures.
The performance results are consistent. There are comparative results. Figures and tables in the paper are related and explained in the relevant paragraph.
Future works have been given in the conclusion.
The references are appropriate.
The conclusions are consistent with the evidence and arguments presented.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
It is interesting study. However, it would be helpful to provide more details on the potential impact of using AI in agriculture beyond reducing labor costs and promoting efficient planting especially in abstract. Additionally, it would be beneficial to provide more context on the significance of radish growth stage recognition and how it fits into the larger field of intelligent agriculture. Finally, it comparison with other published articles is questionable. Language and figures must be more optimized.
English language needs minor revisions.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 5 Report
I have reviewed the article and provided suggestions to improve the quality and soundness of the research. Below are comments provided for authors to enhance the effectiveness of the presented study.
1. The starting sentence should be rephrased to provide actual meaning. “The development of agriculture is a field of great importance all over the world. The growth of the world population increases the demand for agricultural output, which brings more and more pressure to the agriculture of all countries.
2. The results “Based on Inception-v3, we proposed 3 improved models and the test accuracy of Oxford flower dataset reached 99.3%, 2.5% higher than before. Additionally, the accuracy of pest and disease dataset also had perfect performance, the accuracy reached 95.6%, 3.3%higher than before. The accuracy of 99.3 % is questionable. How are the results validated?
3. An explanation of Figure 10 and Table 6 is missing. Explain these in a paragraph.
4. Why is cherry radish taken as a research object in this research? What are the main reasons behind selecting this data set?
5. Spaces between in-text citations are missing.
6. Describe various applications of DCGAN in this research study.
7. The results and future recommendations of the study should be mentioned in the conclusion section.
Remove Gramatical and English typos errors.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Thank you for addressing my recommendations. I am satisfied with the revision.
Please check the whole paper for grammar corrections as the paper has some grammatical glitches.
Reviewer 4 Report
Accept.
Reviewer 5 Report
N/A
The English mistakes must be removed from final draft
