Online System for Monitoring the Degree of Fermentation of Oolong Tea Using Integrated Visible–Near-Infrared Spectroscopy and Image-Processing Technologies

Response to Reviewer 1 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes/Can be improved/Must be improved/Not applicable

Are all the cited references relevant to the research?

Yes/Can be improved/Must be improved/Not applicable

Is the research design appropriate?

Yes/Can be improved/Must be improved/Not applicable

Are the methods adequately described?

Yes/Can be improved/Must be improved/Not applicable

Are the results clearly presented?

Yes/Can be improved/Must be improved/Not applicable

Are the conclusions supported by the results?

Yes/Can be improved/Must be improved/Not applicable

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: did you try to use the techniques separately? The effort of datafusion of two techniques looks very large and expensive. NIR alone might be suitable for the qualitative research question?

Response 1:

Thank you for pointing this out. I have tried to use these two methods separately, and I have added the results of using them separately in lines 346-364 of the revised manuscript

Comments 2:  line 75: please provide the botanical name of the variety

Response 2: Iron Goddess Tea.  Tieguanyin, literally meaning "iron goddess" in Mandarin Chinese, is a particular and popular variety of Chinese tea.

Comments 3:   line 90: is this a law? provide reference

Response 3: Thank you for pointing this out. This study divides the degree of fermentation according to this tradition.

Comments 4:   Line 91: how does sampling every 10 minutes over 6.5 hours lead only to 7 samples                                                           

Response 4:  Here it has been modified to: Sample collection every 10 minutes during fermentation, with 7 samples collected at a time.  Line 91

Comments 5: Line 92/93: it is unclear what are batches and how the resulting samples numbers are occurring

Response 5: I have made corrections here, on lines 91-93 of the revised manuscript.

Comments 6:  line 99: provide references for the practices of "different researchers"

Response 6: This is to explain the manual error, the revised content in the revised draft 99-101lines.

Comments 7: section 2: please add a section about software suppliers, version and algorithm references if needed. Please also add a section about method validation approach (this is currently stated in the results, please move and merge)

Response 7:Agree. I added the validation method section and attached the formula to the revised manuscript on lines 246-264

Comments 8: line 257 and throughout: please provide city, country for all suppliers

Response 8: Thank you, I have finished adding .line 277

Comments 9: line 339: why was only 1 PC used. This is highly unusual for PCA/LDA

Response9:According to the actual data analysis, the first principal component accounts for 96% of the variable feature expression in the LDA, which means that it contains almost all the discriminant information in the data. In this case, it is reasonable to choose a principal component as a discriminating feature because this component is already strong enough to delineate the major differences between the categories. In addition, the adoption of a single principal component can significantly simplify the complexity of the model, reduce computational costs, and improve the interpretability of the model. Although the data involves three categories, two discriminant components may normally be required to describe a complete separation between categories, but since the first principal component already contains the vast majority of the category-distinguishing information, it is sufficient for efficient classification.

In a multi-class problem, reliance on a single principal component can lead to the risk of oversimplifying the data structure. A single component may not fully capture the complex relationships between all categories, especially when the differences between the categories are more than just in the most dominant direction. As a linear technique, LDA is mainly suitable for cases where data types can be clearly separated by linear boundaries. In practice, if the distribution of the data is non-linear, LDA may not be able to effectively distinguish between all categories.

Therefore, further use of nonlinear pattern recognition analysis methods to process data, such as support vector machines (SVM), BPANN, etc., can better deal with nonlinear relationships and complex patterns in data, and can more deeply mine data characteristics to capture details that simple LDA may ignore. Therefore, the generalization ability and classification accuracy of the model on diverse data are improved.

Table S2 is Eigenvalue and cumulate contribution rates of the principal components.

Comments 10:  Table 1: what are the values in the table. Provide a descriptive legend

Response10: Thank you. I've finished it

Comments11: Table 2: perhaps specify in the methods section how you calculate specificity, accuracy and error

Response 11: Finished. Line261-264

4. Response to Comments on the Quality of English Language

Point 1:line 116: two lamps

Response 1:   line 117 (in red)

Point 2: line 280: darkened of the color?

Response 2:   line 300 (in red)

Point 3:line 351: was achieve?

Response 3:   line 387 (in red)

5. Additional clarifications

Principal Component

Eigenvalue

Percentage of Variance (%)

Cumulative (%)

1^st

10.590

96.281

96.281

2^nd

0.347

3.155

99.437

3^rd

0.038

0.347

99.784

4^th

0.013

0.126

99.910

5^th

0.005

0.051

99.962

6^th

0.002

0.025

99.987

7^th

6.96298E-4

0.006

99.993

8^th

4.04615E-4

0.003

99.997

9^th

2.13934E-4

0.002

99.999

10^th

5.27436E-5

4.79488E-4

99.999

Response to Reviewer 2

Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes/Can be improved/Must be improved/Not applicable

Are all the cited references relevant to the research?

Yes/Can be improved/Must be improved/Not applicable

Is the research design appropriate?

Yes/Can be improved/Must be improved/Not applicable

Are the methods adequately described?

Yes/Can be improved/Must be improved/Not applicable

Are the results clearly presented?

Yes/Can be improved/Must be improved/Not applicable

Are the conclusions supported by the results?

Yes/Can be improved/Must be improved/Not applicable

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The title should be changed as the spectral band it uses is visible wavelengths.

Response 1: Thank you for pointing out this problem, I have revised the title

Comments 2: Line 82: How much of the sample was added to the fermenter?

Response 2: I added the sample weight to the revised manuscript. Line79

Comments 3:  Line 90: Law or tradition?. Replace law by tradition.

Response 3: Thank you for pointing this out. I have replaced law with tradition.line90

Comments 4:  Lines 90-93: Not understood. Describe better.                                                  

Response 4:  I have modified this part of the statement. Line88-93

Comments 5: Line 107: Does it fall directly from the fermentation chamber? How do the fermentation chamber and the carrier belt communicate?

Response 5: The conveyor belt is installed under the fermentation chamber, and the tea leaves can cover the conveyor belt.

Comments 6: Line 149: The range 400-950 nm, includes the visible.

Response 6:Thank you, I have replaced all NIR with vis-NIR

Comments 7: Line 151: Include vis-NIR

Response 7:Thank you,  I have replaced all NIR with vis-NIR

Comments 8: Line 151: If the range from 400 to 950 has 550 wavelengths, where does the data of 1517 wavelengths come from? What is the resolution of the equipment?

Response 8: Spectral data was collected in the wavelength range of 400-950 nm at a spectral interval of approximately 0.42, containing 1517 variables.

Comments 9: Lines 209-213: The reference is not included at the end of the document. It is indicated twice in the same paragraph. Indicate only once.

Response9: I've added this quote.

Comments 10:  Paragraph 3.1: The conveyor belt is still subjected to a cleaning process between each tea sample?

Response10: Since the tea will completely cover the conveyor belt, the cleanliness of the conveyor belt will not affect the collection of data, so there is no need to clean.

Comments11:  Lines 292-293: These three wavelengths correspond to the visible. The 450 to the blue, 690 and 770 to the visible red. Only 900 corresponds to the near infrared. The title of the article should be changed as it is misleading.

Response 11: Thank you for pointing out this problem, I have revised the title