Identification of Rice Freshness Using Terahertz Imaging and Deep Learning

Round 1

Reviewer 1 Report

This manuscript analyzed the different freshness levels of rice by using THz reflection imaging. The proposed deep learning network, 1D-VGG19-Inception-ResNet-v2, was used to recognize the freshness levels of the rice. The results demonstrate the feasibility of detecting rice freshness using THz imaging and deep learning. Generally, the manuscript is well written. However, there are some issues in the form as well as in the scientific content. The authors are advised to address them to meet the publication requirements.

1.The storage temperature has a significant effect on rice freshness. The three different temperatures were selected in the manuscript. Please give the reason. And the table 1, which needs to be further supplemented.

2.In the Discussion section, the references of 26-28 should be deleted.

3. The characters in Figure 12 to 15 is too small. It is recommended to enlarge the font size so that the reader can see it clearly.

4. There is too much noise in Figure 13 and 14, which needs further denoising treatment, such as smooth.

4. The Conclusion section is too long. Please condense the conclusion.

The manuscript can be published after revised the issues.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The work was intended to characterize the freshness of rice in reference to storage temperature using THz imaging.

I went through the manuscript, saw different methods and algorism used in an endeavor at characterizing the rice grains freshness, but from my point of view, specific features characterizing a fresh rice are lacking. The work is looking for rice freshness but there isn’t a clear definition of a fresh rice to guide the study.

I placed some questions and suggestions throughout the manuscript to the authors.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors showed that the proposed 1D-VGG19-Inception-ResNet-A network is effective in identifying the freshness of rice by analyzing terahertz reflection image signals with high accuracy and in a relatively short time.

 

1. As an example of the effectiveness of this network, a 2D reflection image in a 50x50mm2 area was analyzed using a terahertz pulse signal. In fact, since freshness is affected by temperature, and freshness affects the reflectance of the terahertz signal, it seems that the correlation between reflectance and freshness can be obtained simply in a much shorter time without using this method, What are the advantages of using this method?

2. As a result, how about another example of the validity of the proposed network? Realistically, in order to obtain a two-dimensional image using a terahertz pulse signal, more time and more sophisticated sample preparation are required compared to conventional near-infrared spectroscopy, electronic nose, gas chromatography, and mass spectroscopy. It seems.. Please compare in a little more detail in this regard.

3. Additionally, as a method to show validity, prepare random samples of unknown temperature and storage time, such as a blind test, and compare and analyze the proposed network with the data analyzed by the existing orthodox method.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

First of all, I would like to thank you for giving me the opportunity to review this paper. The author's diligent efforts and passion for research are well reflected in the paper. We look forward to the results of future research.

I hope that you will continue to show the passion and insight shown in this paper in your next research, and look forward to further development of research in the future.