Diversity of Plum Stones Based on Image Texture Parameters and Machine Learning Algorithms

Round 1

Reviewer 1 Report

The article investigated the applicability and performance of image processing combined with machine learning techniques to distinguish plum stones by their varieties. The development of fast, non-destructive and even contactless methods for the authentication and characterization of plum seeds of different cultivars can be of interest both for economical and quality assurance reasons. The author has used this approach in a unique way, and has not been earlier reported in the literature.

The Introduction section correctly situates the issue, summarizes and cites relevant literature. However, several doubts arose about the implementation of the objectives formulated, especially in the “Materials and Methods” and “Results” sections. It is principally questionable what parameters were used for the modeling, whether and how the models were tested. Therefore, the materials and methods section has to be greatly improved (e.g. nr of samples, applied validation etc.). You can find below my remarks and questions. 

Was there any particular reason why these examinations were conducted on kernels and not on whole intact plums?

Line 16: “The highest results” results should not be named high.

Lines 68-70: When articulating the aims of the research, it was written that "The objective of this study was to evaluate the usefulness of machine learning approach based on image textures selected from a set of about two hundred parameters…" what parameters do you mean here exactly?

Lines 70-72: You mention here and later as well the application of certain performance metrics (Precision, MSS, F-Measure) … how do they contribute to the interpretation of the results? What do they cover?

How many plum stones per variety were tested and how many times were imaged? Please describe what do you mean by image texture, the R, G, B, etc. values per pixel or per plum stone? How does the "around 2000" calculation come out?

In section “2.3. Discrimination” (which would deserve a more descriptive title), it is written that machine learning, discriminant analysis was applied. This is known to be a multivariate statistical analysis, so it is questionable how and why you applied it when you evaluated the results of each color channel separately, since there was only one variable in these cases. Discriminant analysis can be applied to at least two texture parameters simultaneously. This means that there is no point in running analyses for a single-color channel, could instead be analyzed by color spaces (RGB, Lab, XYZ).

Lines 129-130: What do they mean, what they are for and how the indicators mentioned here are calculated? What values can they take?

Lines 131-132: The abbreviations mentioned here refer to exactly what kind of analysis (full name required)? Specify the principle on which the classification is made.

These above-listed questions could be avoided if the respected explanation is given in the materials and methods section.

In lines 62-63 of the “Introduction” and in lines 275-282 of the “Discussion”, as if you were referring to training and predictive modeling, as well as the testing of algorithms (line 127). How did you test the algorithms/models you used? For classification studies, it is important to validate the constructed models to determine the performance of our developed method for unknown samples. Include details.

In Table 1., the classification accuracies and performance metrics are summarized of models including textures selected from color channels. However, no details on which variables (color channels) contributed most to the accuracy of the models. This would be advisable to include in the table or in the text body.

In Figure 4., on what basis did you choose color channels G, L, X, B for the illustration? It may also be advisable to run an analysis of variance on these results to detect significant differences.

It is necessary to clarify previously (“Materials and Methods” section) that what is meant by selected textures, because LDA or QDA cannot be performed only on variable parameter (i. e., color channel). If this is the case, the models need to be recalculated with more variables, besides, the validation results must also be reported.

Author Response

The article investigated the applicability and performance of image processing combined with machine learning techniques to distinguish plum stones by their varieties. The development of fast, non-destructive and even contactless methods for the authentication and characterization of plum seeds of different cultivars can be of interest both for economical and quality assurance reasons. The author has used this approach in a unique way, and has not been earlier reported in the literature.
Answer: Thank you very much for your careful reading and reviewing of the manuscript.

The Introduction section correctly situates the issue, summarizes and cites relevant literature. However, several doubts arose about the implementation of the objectives formulated, especially in the “Materials and Methods” and “Results” sections. It is principally questionable what parameters were used for the modeling, whether and how the models were tested. Therefore, the materials and methods section has to be greatly improved (e.g. nr of samples, applied validation etc.). You can find below my remarks and questions.
Answer:
Thank you for these valuable comments. All the suggestions of the Reviewer have been included in the revised manuscript to further improve it. The Introduction has been extended to include a research justification and a more detailed indication of the novelty of the research. The materials and methods section has been supplemented. Also, the description of the Results has been improved. Detailed information is available in the responses to the comments below.

Was there any particular reason why these examinations were conducted on kernels and not on whole intact plums?
Answer:
The objective of this study was not to distinguish the cultivars of plum fruit. The research focused only on plum stones that can be by-products during plum processing. Therefore, plum stones can be used in industry independently from fruit. Plum stones are considered as products for further processing. The developed procedure can be used for plum stones that have already been extracted from fruits during processing.
It has been indicated in the text as follows:
“During plum processing, flesh and skin are the main raw materials and stones are generated as by-products [9]. Stone of plum consists of the seed covered with a hull. The seed may be a source of useful substances for food, cosmetic (e.g., personal care products) and pharmaceutical industries [10,11]. Plum seeds have a high content of proteins and lipids, but also amygdalin [10,12,13]. The cultivar affects the properties of plum stones and kernels and thus the quality of their products [9]. Therefore, the correct cultivar recognition of plum stones previously extracted from the fruit is important before their further processing.” (lines 45-52)
The plum fruit cultivars were not distinguished. The fruits were not taken into account at all. The image analysis focused on plum stones previously extracted from the fruit. The production of a large amount of waste in the form of fruit stones is often found in the processing of the drupe. The cultivar of plum stones or kernels should be correctly determined so that these by-products can be properly used in further processing.
The plum stones were not damaged during the image analysis (“the undamaged stones were subjected to imaging.” (line 88)). Therefore, the procedure of distinguishing plum stone cultivars is non-destructive.
As indicated in lines 53-56: “The application of machine learning (ML) algorithms may be useful to carry out the discrimination of samples based on the texture, color and shape
parameters extracted from digital images. Due to this, the analysis is performed in a non-destructive, quick, in-expensive, reliably and objective way [14].”.
Therefore, the image acquisition involved only placing undamaged plum stones on the background and taking the images. It has been mentioned in lines 91-94 as follows:
“The whole plum stones were imaged using a digital camera placed on a tripod, LED illumination and a black box. Images were acquired under room conditions. Color calibration of the digital camera was carried out. The stones were positioned on a black background. This facilitated the segmentation of the images.”

Line 16: “The highest results” results should not be named high.
Answer: It has been corrected according to the Reviewer’s comment and changed into “the most satisfactory results” (lines 16-17, 272)

Lines 68-70: When articulating the aims of the research, it was written that "The objective of this study was to evaluate the usefulness of machine learning approach based on image textures selected from a set of about two hundred parameters…" what parameters do you mean here exactly?
Answer: The sentences including the objective have been rewritten as follows:
“The objective of this study was to evaluate the usefulness of machine learning approach based on texture parameters selected from a set of about 1600 textures extracted from images color converted to color channels R, G, B, L, a, b, X, Y, Z to discriminate the plum stone cultivars. The criteria considered for evaluating discrimination were accuracies and the values of other performance metrics such as Precision, MCC (Matthews Correlation Coefficient), F-Measure, Kappa statistic, mean absolute error and root mean squared error. The available literature lacks information on the use of such a large data set (1600 attributes) including textures from individual color channels of images and machine learning algorithms from different groups (Bayes, Lazy, Trees, Meta, Functions, Rules) for developing models for the evaluation of plum stone cultivar diversity that is a great novelty of the present study. The innovation of the study involving developing models for textures selected from a set including textures from all color channels R, G, B, L, a, b, X, Y, Z (about 1600 textures), as well as separately for each color channel (about 180 textures for each of nine color channels).” (lines 69-82)

Lines 70-72: You mention here and later as well the application of certain performance metrics (Precision, MSS, F-Measure) … how do they contribute to the interpretation of the results? What do they cover?
Answer:
These metrics were computed based on the values of TP: True Positive, TN: True Negative, FP: False Positive and FN: False Negative. The metrics (Precision, MCC, F-Measure) provide additional information about the classification performance. The higher the values of the indicators, the more correct the classifications are. It has been specified as follows:
“The discriminant analysis allowed for the determination of the confusion matrices with discrimination accuracies as well as the values of Precision, MCC (Matthews Correlation Coefficient) and F-Measure. The equations for the computation of these metrics are presented in Equations 1-4. The range of values for accuracy was 0-100%, and for Precision, MCC and F-Measure – 0.000-1.000. The decision on the effectiveness of the model was made based on the highest performance metrics. The higher the values of the metrics, the more correct the classifications are.” (lines 154-160)
????????=??+????+??+??+???100
(1)
?????????=????+??
(2)
???=(??∗??)−(??∗??)√(??+??)∗(??+??)∗(??+??)∗(??+??)
(3)
?1−???????=2??2??+??+??
(4)
TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative

How many plum stones per variety were tested and how many times were imaged? Please describe what do you mean by image texture, the R, G, B, etc. values per pixel or per plum stone? How does the "around 2000" calculation come out?
Answer:
It has been specified as follows:
“In the case of each cultivar ‘Emper’, ‘Kalipso’ and ‘Polinka’, one hundred imaged stones were obtained. Ten images were taken for each cultivar. There were ten plum stones in each image. In total, the obtained set consisted of images for three hundred plum stones.” (lines 94-97)
“For each plum stone considered as a set of pixels, the region of interest (ROI) was overlaid. The image textures were considered as a function of the spatial variation of the pixel brightness intensity. The images can have repeated subpatterns of distribution of pixel brightness. Textures can provide information about the structure of the objects. The quantitative texture analysis can provide important insights about object quality [19]. The texture parameters were calculated for stone images converted to color channels R, G, B, L, a, b, X, Y, Z. (lines 102-108)
“In the case of each ROI (each plum stone) in each of nine color channels, about 180 textures based on the co-occurrence matrix (132 texture parameters including 11 features computed for 4 various directions and 3 between-pixels distances), run-length matrix (20 texture parameters including 5 features computed for 4 various directions), histogram (9 texture parameters), Haar wavelet transform (10 texture parameters), gradient map (5 texture parameters), and autoregressive model (5 texture parameters) were computed [27]. In total, about 1600 image textures were determined for each stone. These textures were used first for the selection step and then for the development of models to distinguish plum stone cultivars.” (lines 113-121)

In section “2.3. Discrimination” (which would deserve a more descriptive title), it is written that machine learning, discriminant analysis was applied. This is known to be a multivariate statistical analysis, so it is questionable how and why you applied it when you evaluated the results of each color channel separately, since there was only one variable in these cases.
Answer:
The title for the subsection has been changed into: 2.3. Plum stone cultivar discrimination using machine learning algorithms.
For each color channel, about 180 variables (textures) have been determined. The procedure has been specified as in the answer to the comment above:
“In the case of each ROI (each plum stone) in each of nine color channels, about 180 textures based on the co-occurrence matrix (132 texture parameters including 11 features computed for 4 various directions and 3 between-pixels distances), run-length matrix (20 texture parameters including 5 features computed for 4 various directions), histogram (9 texture parameters), Haar wavelet transform (10 texture parameters), gradient map (5 texture parameters), and autoregressive model (5 texture parameters) were computed [27]. In total, about 1600 image
textures were determined for each stone. These textures were used first for the selection step and then for the development of models to distinguish plum stone cultivars.” (lines 113-121)

Discriminant analysis can be applied to at least two texture parameters simultaneously. This means that there is no point in running analyses for a single-color channel, could instead be analyzed by color spaces (RGB, Lab, XYZ).
Answer:
As indicated in the comments above, for each of nine color channels, about 180 textures have been determined.

Lines 129-130: What do they mean, what they are for and how the indicators mentioned here are calculated? What values can they take?
Answer:
It has been specified as follows:
“The discriminant analysis allowed for the determination of the confusion matrices with discrimination accuracies as well as the values of Precision, MCC (Matthews Correlation Coefficient) and F-Measure. The equations for the computation of these metrics are presented in Equations 1-4. The range of values for accuracy was 0-100%, and for Precision, MCC and F-Measure – 0.000-1.000. The decision on the effectiveness of the model was made based on the highest performance metrics. The higher the values of the metrics, the more correct the classifications are.” (lines 154-160)
????????=??+????+??+??+???100
(1)
?????????=????+??
(2)
???=(??∗??)−(??∗??)√(??+??)∗(??+??)∗(??+??)∗(??+??)
(3)
?1−???????=2??2??+??+??
(4)
TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative

Lines 131-132: The abbreviations mentioned here refer to exactly what kind of analysis (full name required)? Specify the principle on which the classification is made.
Answer:
It has been corrected as follows:
“IBk (Instance-Based Learning with parameter k) (Lazy), QDA (Quadratic Discriminant Analysis) (Functions), LDA (Linear Discriminant Analysis) (Functions) and Random Forest (Trees)” (lines 164-165)
“The classification was performed using a 10-fold cross-validation mode. The dataset was randomly divided into 10 parts. The learning was performed a total of 10 times using different training sets. Each of 10 parts was used as the test set, and the remaining 9 parts - as the training sets in turn. The results were the average of 10 estimates [30].” (lines 150-154)
“Based on the most successful values of metrics for obtained the IBk (Instance-Based Learning with parameter k) (Lazy), QDA (Quadratic Discriminant Analysis) (Functions), LDA (Linear Discriminant Analysis) (Functions) and Random Forest (Trees), the results for these algorithms were chosen to be presented in this paper. The parameters of these selected algorithms were the following: for IBk - doNotCheckCapabilities: False, batchSize: 100, KNN: 1, debug: False,
meanSquared: False, windowSize: 0, nearestNeighbourSearchAlgorithm: LinearNNSearch – distanceFunction: EuclideanDistance -R first-last; for QDA and LDA - doNotCheckCapabilities: False, ridge: 1.0E-6, batchSize: 100, debug: False; for Random Forest - bagSizePercent: 100, batchSize: 100, breakTiesRandomly: False, debug: False, doNotCheckCapabilities: False, numExcecutionSlots: 1, numInteractions: 100, seed: 1.” (lines 163-173)

These above-listed questions could be avoided if the respected explanation is given in the materials and methods section.
Answer:
The Materials and Methods section has been extensively supplemented (please see the answer to comments above).

In lines 62-63 of the “Introduction” and in lines 275-282 of the “Discussion”, as if you were referring to training and predictive modeling, as well as the testing of algorithms (line 127). How did you test the algorithms/models you used? For classification studies, it is important to validate the constructed models to determine the performance of our developed method for unknown samples. Include details.
Answer:
The validations for unknown samples have been performed. More details have been added to the manuscript:
“The classification was performed using a 10-fold cross-validation mode. The dataset was randomly divided into 10 parts. The learning was performed a total of 10 times using different training sets. Each of 10 parts was used as the test set, and the remaining 9 parts - as the training sets in turn. The results were the average of 10 estimates [30].” (lines 150-154)

In Table 1., the classification accuracies and performance metrics are summarized of models including textures selected from color channels. However, no details on which variables (color channels) contributed most to the accuracy of the models. This would be advisable to include in the table or in the text body.
Answer:
It has been indicated in the text as follows:
“The textures contributing most to the accuracy of the models were from color channels G and B. They were characterized by the highest discriminative power. Examples of these textures are GS5SH5SumOfSqs, GS5SV1SumVarnc, BS5SZ1SumOfSqs, BS5SV1SumOfSqs, GHPerc10, GHKurtosis, BHPerc10.” (lines 206-210)

In Figure 4., on what basis did you choose color channels G, L, X, B for the illustration? It may also be advisable to run an analysis of variance on these results to detect significant differences.
Answer:
It has been specified as follows:
“The examples of original plum stone images and images converted to randomly selected color channels for which significant differences between HMean textures were determined based on the results performed using a one-way ANOVA in STATISTICA (StatSoft Inc., Tulsa, USA) software at a significance level of p ≤ 0.05, are shown in Figure 2.” (lines 206-210)
“The results of a one-way ANOVA revealed that differences in means of GHMean, LHMean, XHMean and BHMean between ‘Emper’, ‘Kalipso’, and ‘Polinka’ plum stones were statistically significant, and each cultivar formed the separate homogenous group. The values of GHMean were equal to 140.34 for ‘Polinka’, 160.60 for ‘Emper’ and 170.02 for ‘Kalipso’.
LHMean variable was characterized by the values of 172.97 for ‘Polinka’, 188.15 for ‘Emper’ and 195.12 for ‘Kalipso’. The values of XHMean were equal to 91.56 for ‘Polinka’, 111.65 for ‘Emper’, 121.46 for ‘Kalipso’. The determined BHMean values were equal to 84.21 for ‘Polinka’, 105.05 for ‘Emper’ and 111.12 for ‘Kalipso’.“ (lines 253-261)

It is necessary to clarify previously (“Materials and Methods” section) that what is meant by selected textures, because LDA or QDA cannot be performed only on variable parameter (i. e., color channel). If this is the case, the models need to be recalculated with more variables, besides, the validation results must also be reported.
Answer:
There were a lot of variables. It has been specified as follows:
“In the case of each ROI (each plum stone) in each of nine color channels, about 180 textures based on the co-occurrence matrix (132 texture parameters including 11 features computed for 4 various directions and 3 between-pixels distances), run-length matrix (20 texture parameters including 5 features computed for 4 various directions), histogram (9 texture parameters), Haar wavelet transform (10 texture parameters), gradient map (5 texture parameters), and autoregressive model (5 texture parameters) were computed [27]. In total, about 1600 image textures were determined for each stone. These textures were used first for the selection step and then for the development of models to distinguish plum stone cultivars.” (lines 113-121)
Each model was built based on more than 20 variables.

Author Response File: Author Response.pdf

Reviewer 2 Report

The objective of this study was to evaluate the usefulness of machine learning based on image texture parameters to discriminate the plum stone cultivars.

This is an extensive research, with a lot of numerical analysis. 
Thematically the work is interesting for the researchers and professionals and the proposed manuscript is relevant to the scope of the journal.

Some modifications and clarification are necessary.

The overall organization and structure of the manuscript are appropriate. The paper is well written and the topic is appropriate for the journal.
The aim of the paper is well described and the discussion was well approached, its results and discussion are correlated to the cited literature data.
The literature review is comprehensive and properly done.

The novelty of the work must be more clearly demonstrated.
The significance of the Work: Given the large number of analyzed data, this is an interesting study with a possible significant impact in this area.

Statistical interpretation of the analytical data must be more properly presented.

The verification of the model should be performed. 

More details and more results should be presented in the text?

Other Specific Comments: The work is properly presented in terms of the language. The work presented here is very interesting and well done, it is presented in a compact manner.

Author Response

The objective of this study was to evaluate the usefulness of machine learning based on image texture parameters to discriminate the plum stone cultivars.

This is an extensive research, with a lot of numerical analysis.

Thematically the work is interesting for the researchers and professionals and the proposed manuscript is relevant to the scope of the journal.

Answer:
Thank you very much for reviewing the manuscript and this comment.

Some modifications and clarification are necessary.

Answer:

Thank you very much for this comment. The manuscript has been revised according to all suggestions.

The overall organization and structure of the manuscript are appropriate. The paper is well written and the topic is appropriate for the journal.

The aim of the paper is well described and the discussion was well approached, its results and discussion are correlated to the cited literature data.

The literature review is comprehensive and properly done.

Answer:
Thank you very much for all your comments. To improve the manuscript, all the suggestions have been included in the revised version.

The novelty of the work must be more clearly demonstrated.

Answer:
The novelty of this study has been indicated. It has been specified in the paragraph including the objective of the study as follows:

“The objective of this study was to evaluate the usefulness of machine learning approach based on texture parameters selected from a set of about 1600 textures extracted from images color converted to color channels R, G, B, L, a, b, X, Y, Z to discriminate the plum stone cultivars. The criteria considered for evaluating discrimination were accuracies and the values of other performance metrics such as Precision, MCC (Matthews Correlation Coefficient), F-Measure, Kappa statistic, mean absolute error and root mean squared error. The available literature lacks information on the use of such a large data set (1600 attributes) including textures from individual color channels of images and machine learning algorithms from different groups (Bayes, Lazy, Trees, Meta, Functions, Rules) for developing models for the evaluation of plum stone cultivar diversity that is a great novelty of the present study. The innovation of the study involving developing models for textures selected from a set including textures from all color channels R, G, B, L, a, b, X, Y, Z (about 1600 textures), as well as separately for each color channel (about 180 textures for each of nine color channels). (lines 69-82)

 

 

The significance of the Work: Given the large number of analyzed data, this is an interesting study with a possible significant impact in this area.

Answer:

Thank you very much for all your comments. Information about analyzed data has been more detailed as follows:

“In the case of each cultivar ‘Emper’, ‘Kalipso’ and ‘Polinka’, one hundred imaged stones were obtained. Ten images were taken for each cultivar. There were ten plum stones in each image. In total, the obtained set consisted of images for three hundred plum stones.” (lines 94-97)

“In total, about 1600 image textures were determined for each stone. These textures were used first for the selection step and then for the development of models to distinguish plum stone cultivars.” (lines 119-121)

 
Statistical interpretation of the analytical data must be more properly presented.

Answer:

It has been corrected. Sections 2. Materials and Methods as well as 3. Results have been supplemented. Additional metrics are presented. Tables 1-4 have been supplemented.

The equations for the computation of the accuracy, Precision, MCC (Matthews Correlation Coefficient) and F-Measure have been added in Equations 1-4.

 

 

	(1)
	(2)
	(3)
	(4)

 

TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative

 

The text of the 2. Materials and Methods section has been supplemented with the following sentences:

“The discriminant analysis allowed for the determination of the confusion matrices with discrimination accuracies as well as the values of Precision, MCC (Matthews Correlation Coefficient) and F-Measure. The equations for the computation of these metrics are presented in Equations 1-4. The range of values for accuracy was 0-100%, and for Precision, MCC and F-Measure – 0.000-1.000. The decision on the effectiveness of the model was made based on the highest performance metrics. The higher the values of the metrics, the more correct the classifications are. Additionally, the values of Kappa statistic, mean absolute error and root mean squared error (range of 0.000-1.000) were determined. The high value of the Kappa statistic was considered desirable. Whereas the values of errors should be as low as possible. Based on the most successful values of metrics for obtained the IBk (Instance-Based Learning with parameter k) (Lazy), QDA (Quadratic Discriminant Analysis) (Functions), LDA (Linear Discriminant Analysis) (Functions) and Random Forest (Trees), the results for these algorithms were chosen to be presented in this paper. The parameters of these selected algorithms were the following: for IBk - doNotCheckCapabilities: False, batchSize: 100, KNN: 1, debug: False, meanSquared: False, windowSize: 0, nearestNeighbourSearchAlgorithm: LinearNNSearch – distanceFunction: EuclideanDistance -R first-last; for QDA and LDA - doNotCheckCapabilities: False, ridge: 1.0E-6, batchSize: 100, debug: False; for Random Forest - bagSizePercent: 100, batchSize: 100, breakTiesRandomly: False, debug: False, doNotCheckCapabilities: False, numExcecutionSlots: 1, numInteractions: 100, seed: 1.”

The results have been described in more detail in the 3. Results section.

 

 

The verification of the model should be performed. 

Answer:

The 2. Materials and Methods section has been supplemented as follows:

“To verify the correctness of the approach used automatic General Discriminant Analysis (GDA) model involving the progressive discrimination was tested in STATISTICA (StatSoft Inc., Tulsa, USA) software at suggested parameters. Different software and a different procedure for building the model than in WEKA were chosen to ensure the independence of the analysis. The verification was used for models including textures selected from color channels R, G, B, L, a, b, X, Y, Z built in Weka using various machine learning algorithms.” (lines 197-203)

The 3. Results section has been supplemented as follows:

“The verification of these models built based on textures selected from color channels R, G, B, L, a, b, X, Y, Z (Table 1) carried out using automatic General Discriminant Analysis (GDA) proved the high accuracy of discrimination. The plum stones were correctly discriminated in 96.67%. The accuracies for stones belonging to individual cultivar were equal to 96% for ‘Emper’ and ‘Polinka’ and 98% for ‘Kalipso’.” (lines 237-241)

 

 

More details and more results should be presented in the text?

Answer:

It has been corrected. New sentences have been added as follow:

“Additionally, the values of Kappa statistic, mean absolute error and root mean squared error (range of 0.000-1.000) were determined. The high value of the Kappa statistic was considered desirable. Whereas the values of errors should be as low as possible.” (lines 160-163)

 

Tables 1-4 have been supplemented. The results are described in more detail. Section 3. Results has been supplemented with the following sentences:

“The textures contributing most to the accuracy of the models were from color channels G and B. They were characterized by the highest discriminative power. Examples of these textures are GS5SH5SumOfSqs, GS5SV1SumVarnc, BS5SZ1SumOfSqs, BS5SV1SumOfSqs, GHPerc10, GHKurtosis, BHPerc10.” (lines 206-210)

 

“Also, the Kappa statistic equal to 0.95 was high. Whereas the values of mean absolute error (0.0247) and root mean squared error (0.1369) were low.” (lines 218-220)

“The ‘Kalipso’ plum stones were characterized by a very high Precision (0.990) for the model built using the QDA algorithm. Whereas the values of MCC (0.978) and F-Measure (0.985) were the most satisfactory for the ‘Kalipso’ cultivar and Random Forest algorithm. The values of Kappa statistic were high for QDA (0.93), LDA (0.925) and Random Forest (0.92). In the case of each classifier, mean absolute error and root mean squared error were low of up to 0.1136 and 0.1918 (Random Forest), respectively. The verification of these models built based on textures selected from color channels R, G, B, L, a, b, X, Y, Z (Table 1) carried out using automatic General Discriminant Analysis (GDA) proved the high accuracy of discrimination. The plum stones were correctly discriminated in 96.67%. The accuracies for stones belonging to individual cultivar were equal to 96% for ‘Emper’ and ‘Polinka’ and 98% for ‘Kalipso’.” (lines 231-241)

 

“The results of a one-way ANOVA revealed that differences in means of GHMean, LHMean, XHMean and BHMean between ‘Emper’, ‘Kalipso’, and ‘Polinka’ plum stones were statistically significant, and each cultivar formed the separate homogenous group. The values of GHMean were equal to 140.34 for ‘Polinka’, 160.60 for ‘Emper’ and 170.02 for ‘Kalipso’. LHMean variable was characterized by the values of 172.97 for ‘Polinka’, 188.15 for ‘Emper’ and 195.12 for ‘Kalipso’. The values of XHMean were equal to 91.56 for ‘Polinka’, 111.65 for ‘Emper’, 121.46 for ‘Kalipso’. The determined BHMean values were equal to 84.21 for ‘Polinka’, 105.05 for ‘Emper’ and 111.12 for ‘Kalipso’.” (lines 253-261)

 

“The Kappa statistic was in the range of 0.82 to 0.92, mean absolute error – from 0.0921 to 0.0373 and root mean squared error – from 0.2350 to 0.1747. The lowest values were obtained for the model developed based on selected image textures from color channel R using the LDA algorithm. The highest values were determined in the case of color channel G and QDA.” (lines 287-291)

 

“The highest Precision equal to 0.979 was found for the ‘Kalipso’, color channels L and QDA. The highest MCC (0.941) and F-Measure (0.961) were determined for the ‘Kalipso’, color channels L and LDA. The model built for textures selected from color channels L using QDA was characterized by the highest Kappa statistic (0.905) and the lowest mean absolute error (0.0423). The lowest root mean squared error was observed for color channels L and LDA. Whereas the highest mean absolute error (0.1575, LDA) and root mean squared error (0.3142, QDA), as well as the lowest Kappa statistic (0.695, LDA), were found in the case of classification performed based on textures selected from color channel b.” (lines 304-312)

 

“The most satisfactory Precision (0.990MCC (0.955), F-Measure (0.969) were obtained for the ‘Kalipso’, color channel X and QDA. The Kappa statistic (0.88) was the highest and the values of mean absolute error (0.0561) and root mean squared error (0.2063) were the lowest in the case of the model built for textures selected from color channel X using QDA. In the case of color channel Z and LDA, the Kappa statistic equal to 0.815 was the lowest and the root mean squared error of 0.2650 was the highest. The highest mean absolute error equal to 0.0924 was observed for color channels X and LDA.” (lines 327-334)

 

 

Other Specific Comments: The work is properly presented in terms of the language. The work presented here is very interesting and well done, it is presented in a compact manner.

Answer:

Thank you very much for your comments.

 

Author Response File: Author Response.pdf