Apricot Stone Classification Using Image Analysis and Machine Learning
Round 1
Reviewer 1 Report
In this manuscript, author used machine learning algorithms for distinguishing different
cultivars of apricot stones was demonstrated. This study investigates the effectiveness of two low-cost colour imaging systems coupled with supervised learning to develop classification models to determine the cultivar of different stones. The results indicated the possibility of distinguishing apricot stone cultivars with high accuracy. Further research with more samples will focus on the application of deep learning. I think the manuscript can be published on Sustainability after major revisions. My comment is as follows:
1. In the chapter on image processing, the description of the image processing process is too general and needs to be described in detail, such as the embodiment of the workload in the processing process, and the necessary links can be illustrated with pictures;
2. Is the angle of data acquisition considered in the chapter on imaging systems, please indicate in the manuscript;
3. In the result section, it is too monotonous to use only graphs to show the test results, for example, table1 and table2 would be more intuitive to use bar charts. Learning relevant data visualization techniques will take your collection to the next level;
4. Model evaluation metrics should be listed in a separate subsection, including their formulae and relevance.
Author Response
Comment: In this manuscript, author used machine learning algorithms for distinguishing different cultivars of apricot stones was demonstrated. This study investigates the effectiveness of two low-cost colour imaging systems coupled with supervised learning to develop classification models to determine the cultivar of different stones. The results indicated the possibility of distinguishing apricot stone cultivars with high accuracy. Further research with more samples will focus on the application of deep learning. I think the manuscript can be published on Sustainability after major revisions. My comment is as follows:
- In the chapter on image processing, the description of the image processing process is too general and needs to be described in detail, such as the embodiment of the workload in the processing process, and the necessary links can be illustrated with pictures;
Answer: Please refer to Figure 3 in the updated manuscript where an example was illustrated for the image processing steps.
- Is the angle of data acquisition considered in the chapter on imaging systems, please indicate in the manuscript;
Answer: Please refer to the updated manuscript (lines 145-147) where is was added that the axis of the camera lens was perpendicularly directed on the surface the samples were placed on.
- In the result section, it is too monotonous to use only graphs to show the test results, for example, table1 and table2 would be more intuitive to use bar charts. Learning relevant data visualization techniques will take your collection to the next level;
Answer: The authors partially agree with the reviewer. While Tables 1-2 are simple to follow, others like 7-10 are more difficult for the reader to follow. However, Tables 3-6 have different parameters related to extracted features and they are not uniform for each colour channels; Thus, it will be more confusing for the readers to follow the extracted graphs. The authors, therefore, converted tables 7-10 to Figures 4-7.
- Model evaluation metrics should be listed in a separate subsection, including their formulae and relevance.
Answer: Please refer to the updated manuscript where section 2.4.2. titled “ Evaluation of classification models” was dedicated for explaining the different criteria and formula for evaluating classification models.
Reviewer 2 Report
There are grammatical mistakes somewhere.
At many places ‘between cultivars’ has been used, its better to write ‘among cultivars’
Line 55: No need to write ‘also’ or rewrite the sentence with proper sense.
Line 61: Please write properly ‘which helps to reduce’
Line 126-128: Please write properly making sense.
Table 5: Please remove full stop at the end of table captions.
Author Response
There are grammatical mistakes somewhere.
At many places ‘between cultivars’ has been used, its better to write ‘among cultivars’
Answer: The authors would like to thank the reviewers and assure that the referred errors have been corrected in the whole manuscript.
Line 55: No need to write ‘also’ or rewrite the sentence with proper sense.
Answer: The authors would like to thank the reviewers and assure that the referred error has been corrected.
Line 61: Please write properly ‘which helps to reduce’
Answer: The authors would like to thank the reviewers and assure that the referred error has been corrected (line 61).
Line 126-128: Please write properly making sense.
Answer: Please refer to the updated manuscript (lines: 127-129) where the referred sentence was modified as follows: “For each cultivar, twenty-five stones were scanned. Figure 1 shows examples of the apricot stones’ images used in the study”.
Table 5: Please remove full stop at the end of table captions.
Answer: Please refer to the updated manuscript where the full stops were deleted from the captions of all tables.
Reviewer 3 Report
Here are my comments related to the manuscript,
Line 78; The sentence “In a study conducted by [16], it was shown that the content of sugars, fatty acids, and organic acids in apricot stones varied between cultivars” should be corrected as sentence “In a study conducted by Farag et al. [16], it was shown that the content of sugars, fatty acids, and organic acids in apricot stones varied between cultivars”
Line 126-127; Can you check the sentence? “Twenty-five 126 stones were obtained for each of the”. It seems not completed.
Author Response
Here are my comments related to the manuscript,
Line 78; The sentence “In a study conducted by [16], it was shown that the content of sugars, fatty acids, and organic acids in apricot stones varied between cultivars” should be corrected as sentence “In a study conducted by Farag et al. [16], it was shown that the content of sugars, fatty acids, and organic acids in apricot stones varied between cultivars”
Answer: Please refer to the updated manuscript (lines: 78-81) where the referred sentence has been modified into: “In a study conducted by Farag et al. [16], it was shown that the content of sugars, fatty acids, and organic acids in apricot stones varied among cultivars.”
Line 126-127; Can you check the sentence? “Twenty-five 126 stones were obtained for each of the”. It seems not completed.
Answer: Please refer to the updated manuscript (lines: 127-129) where the referred sentence was modified as follows: “For each cultivar, twenty-five stones were scanned. Figure 1 shows examples of the apricot stones’ images used in the study”.
Reviewer 4 Report
Image recognition technology is a popular topic in machine learning, similar feature classification applications have been widely used in commercial applications. Using image recognition for apricot stones does not increase the novelty of this study.
Most importantly, there is a major limitation of this study, which is that the data set was too small with 2172 features and only 125 total samples (n/m). This low number of data might lead to a high risk of overfitting and low generalizability. Increased data size would be needed to improve our model accuracy. In addition, it would be helpful to include the result curve for each discussion section.
Author Response
Image recognition technology is a popular topic in machine learning, similar feature classification applications have been widely used in commercial applications. Using image recognition for apricot stones does not increase the novelty of this study.
Most importantly, there is a major limitation of this study, which is that the data set was too small with 2172 features and only 125 total samples (n/m). This low number of data might lead to a high risk of overfitting and low generalizability. Increased data size would be needed to improve our model accuracy. In addition, it would be helpful to include the result curve for each discussion section.
Answer: The authors agree that the data sets in the study are relatively are relatively small. However, the study tested five cultivars which indeed helped generated more generalized classification models. Furthermore, the feature selection algorithms applied in the study yielded a significant reduction of the features from 2172 to a maximum of 23 features as shown in Table 3. Furthermore, the authors stated in the conclusion section (lines: 420-423) that “Further research should consider larger data sets as well as more diverse cultivars which increase the feasibility of applying advanced machine learning techniques especially deep learning and improve the likelihood for transfer learning between different cultivars”.
Round 2
Reviewer 1 Report
I have carefully reviewed your manuscript and I am pleased to inform you that I recommend it for publication in [Journal Name]. The study is well-conducted, and the results are of significant importance to the field. The paper is well-written and easy to understand.
The methodology used is appropriate, and the data analysis is thorough. The conclusion drawn from the study is well-supported by the data. The references cited are relevant and up-to-date.
Overall, I believe that your manuscript makes a valuable contribution to the field and would be of interest to our readers. I recommend that it be accepted for publication without any major revisions.
Thank you for considering my comments.
