Superior EVOO Quality Production: An RGB Sorting Machine for Olive Classification
, Lavinia Moscovini 1, Corrado Costa 1, Paolo Del Re 2, Lucia Giansante 2, Pietro Toscano 3, Francesco Tocci 1, Simone Vasta 1, Rossella Manganiello 1, Luciano Ortenzi 1 and Federico Pallottino 1,*
Round 1
Reviewer 1 Report
* Lines 27 and 131: The claim "“For the first time, the optoelectronic technologies in
an industrial system was tested on olives to produce superior quality 28 EVOO” is not accurate.
See for instance European project :
https://cordis.europa.eu/project/id/811930
* Please review spelling. For instance, in line 198 "invaiate". In line 164 "owing->belonging"?,...
* Table 2: totals are inconsistent. Please review the numbers.
* General comment: the olives categories are not clearly named through the paper.
- At the beginning of Materials and Methods section, they mention ripeness and defects criteria to sort into three categories "bad", "good" and "spot".
- Then, authors talk about classes ("Mix choice", "first choice", "second choice".
- Then in Figure 2 we can see "mixed", "green" and black and in table2 "first class", "second class", and "mixed".
Some sort of unification would be desired.
* Heading of table 4. Review line breaks, for instance EVOOMix_1
* Little detail is given about the computer vision sorting methods.
* General Opinion. The paper is somewhat dense, specially the materials and methods section.
Too many numbers, and sure all of them are relevant for the conclusions of the paper.
The results obtained are not particularly relevant neither in classification efficiency nor in proving the relation of classificacion with quality. The only clear conclusion is that there are slight
differences in organoleptic features, mainly due to:
- Ripness classification is done using color but there is a range of ripeness that can not be
observed by color.
- Ripeness index of olives is not related only with color. When the olive skin is black, there may be different degrees of maturity and an overripe olive is as dark as an olive hat has just turned from green to black. The only way to tell the differences (as far as I know) is to use other wavelengths such as infrared.
- It is very difficult that chemical-physical parameters show a degradation based only on the color classification of input olives. The oil quality depends more on the presence with a high enough percentage of certain defects.
- Another limitation is that defects can only be detected on green olives. Dark defects can not be detected in black olives.
- The efficiency of the classification is not very good as can be seen in the % of errors. This makes that there is a significant mix at the different outputs, and the results can not be se clear.
Author Response
Comments and Suggestions for Authors
1) Lines 27 and 131: The claim “For the first time, the optoelectronic technologies in an industrial system was tested on olives to produce superior quality 28 EVOO” is not accurate. See for instance European project :https://cordis.europa.eu/project/id/811930
In the European Project you pointed out to us, there is no mention of an industrial sorter, but rather of a sorter that could represent a breakthrough in technology but is still patent pending. Instead, the optical sorter used in our work is already for sale and in use in industrial companies.
2) Please review spelling. For instance, in line 198 “invaiate”. In line 164 “owing->belonging”?,...
We reviewed words in the text. We replaced “invaiate” with “ripened” and “owing” with “relating” as suggested to reviewer 3.
3) Table 2: totals are inconsistent. Please review the numbers.
We reviewed and modified the numbers.
4) General comment: the olives categories are not clearly named through the paper.
- At the beginning of Materials and Methods section, they mention ripeness and defects criteria to sort into three categories “bad”, “good” and “spot”.
- Then, authors talk about classes “Mix choice”, “first choice”, “second choice”.
- Then in Figure 2 we can see “mixed”, “green” and black and in table2 “first class”, “second class”, and “mixed”.
- Some sort of unification would be desired.
At the beginning of the M&M section, three criteria are mentioned, namely “Good”, “Bad” and “Spot” which refer to the olive grading method used by the grader during operation. The classes “Mix choice,” “First choice” and “Second choice” correspond to those shown in Figure 2 and Table 2. As you suggested, we have unified the classes all under one name (First choice, Second choice and Mixed choice).
5) Heading of table 4. Review line breaks, for instance EVOOMix_1
Thanks for your comment. We reviewed line breaks.
6) Little detail is given about the computer vision sorting methods. FEDERICO
Additional details are now present in the text.
7) General Opinion. The paper is somewhat dense, specially the materials and methods section. Too many numbers, and sure all of them are relevant for the conclusions of the paper. The results obtained are not particularly relevant neither in classification efficiency nor in proving the relation of classificacion with quality. The only clear conclusion is that there are slight differences in organoleptic features, mainly due to:
- Ripness classification is done using color but there is a range of ripeness that cannot be observed by color.
- Ripeness index of olives is not related only with color. When the olive skin is black, there may be different degrees of maturity and an overripe olive is as dark as an olive hat has just turned from green to black. The only way to tell the differences (as far as I know) is to use other wavelengths such as infrared.
Thanks for the comment we are aware of that and taking this in mind we wanted to verify what was the output of this kind of instruments that are being sold. However, since the machine include an IR camera, we plan to deepen the study including the IR acquisition and processing during the project (this was already present in the conclusion).
- It is very difficult that chemical-physical parameters show a degradation based only on the color classification of input olives. The oil quality depends more on the presence with a high enough percentage of certain defects.
- Another limitation is that defects can only be detected on green olives. Dark defects cannot be detected in black olives.
Again, thank you, this consideration has been added in the conclusion of the manuscript now.
- The efficiency of the classification is not very good as can be seen in the % of errors. This makes that there is a significant mix at the different outputs, and the results cannot be se clear.
Thank you, we do hope to significantly improve the output through the combined use of RGB and infrared sensors, however there are already discrete differences in the final oils evaluated.
Reviewer 2 Report
The olives ripeness and the choice of harvest time according to their color and size, strongly influences the quality of the EVOO. The physical sorting of olives with machines performing rapid and objective optical selection, impossible by hand, can improve the quality of the final product. The authors described that the classification of olives into two qualitative classes, based on the maturity stage and the presence of external defects, through an industrial RGB optical sorting prototype, evaluating its performance and comparing the results with those obtained visually by trained operators. That work is interesting and very useful for olive oil quality-improvement. However, some revisions should be made before publication. They are showed as follows.
Comments:
1) In the part of “Discussion”, the discussion the authors gave is not deep, and is mainly lack of comparison with other reported literatures.
2) Previous literature “Olive classification according to RGB, HSV and L*a*b* color parameters using Image processing technique” have showed that the defected and sound olives by image processing technique could be discriminated. What is new or innovation of your work？please be stated.
3) The description on the cutting-edge technique concerning RGB sorting of this work in Introduction seems not enough.
4) Visual evaluation in Line 196: the sorting rules of olive ripeness (green, mix and black) are not explicit. A quantification rules should be suggested.
Author Response
Comments and Suggestions for Authors
The olives ripeness and the choice of harvest time according to their color and size, strongly influences the quality of the EVOO. The physical sorting of olives with machines performing rapid and objective optical selection, impossible by hand, can improve the quality of the final product. The authors described that the classification of olives into two qualitative classes, based on the maturity stage and the presence of external defects, through an industrial RGB optical sorting prototype, evaluating its performance and comparing the results with those obtained visually by trained operators. That work is interesting and very useful for olive oil quality-improvement. However, some revisions should be made before publication. They are showed as follows.
Comments:
1) In the part of “Discussion”, the discussion the authors gave is not deep, and is mainly lack of comparison with other reported literatures.
The discussion has not been deepened by referring to other publications because there is not much literature on the subject (as cited in the introduction); in particular, the sorters that are used are not industrial as in our work and consequently comparisons cannot be made.
2) Previous literature “Olive classification according to RGB, HSV and L*a*b* color parameters using Image processing technique” have showed that the defected and sound olives by image processing technique could be discriminated. What is new or innovation of your work? please be stated.
The innovations related to this work are not only about sorting olives by discriminating between defective and healthy olives using image processing technique. The work aims to produce a higher quality EVOO by eliminating defective olives from selection. In fact, the panel test performed on “First choice,” “Second choice” and “Mixed choice” olives also reported different results (this topic has already been expressed in the discussion and conclusion).
3) The description on the cutting-edge technique concerning RGB sorting of this work in Introduction seems not enough.
As suggested two references regarding the use of RGB associated with EVOO have been added in the Introduction. Regarding RGB sorting, no more was found other than the one already mentioned in the article, as there are not many works on this subject. Consequently, this work of ours represents a novel approach using an RGB chamber applied to an industrial, non-laboratory sorting machine.
4) Visual evaluation in Line 196: the sorting rules of olive ripeness (green, mix and black) are not explicit. A quantification rules should be suggested.
The quantification rule is expressed in lines 289-294: “The yield of an oil is obtained from the physical extraction of the olives (25 percent of the initial weight, generally). To obtain a higher yield, olives should be harvested when the drupes are ripe and thus capable of giving the best oil yield [24]. In general, for the purposes of oil quality, the optimal harvesting period is strongly influenced by the degree of ripeness of the drupes. Harvesting should be done when the olives are fully versioned and give a high oil yield relative to fresh weight [25].” Given this statement related to the quality of EVOO, three classes were identified based on olive ripeness.
Reviewer 3 Report
Line 36 About 82 %
Is reference 3 correct as I could not find reference to % of production in Spain and Italy. If not please correct.
Line 41 you refer to mechanical methods yet do not list a mechanical method
Line 42 For an EVOO to be defined as such it
Line 45 and the other containing the
Reference 6 - are you sure this is correct. Please go through the first part of the reference section and make sure that the references do indeed say what you quote.
Line 60 in the phase between the green and
Line 62 may vary according to the production region and cultivars milled
Line 63 the genuiness of the
Line 71 allow the determination of the quality of the oil with respect to sensory characteristics
Please check with editor if Food uses US or British spelling. If British then correct colour, flavour
Line 80 ingredients with a strong fruit flavour.
Line 93 oil, panel testing represents a crucial
Line 93 This allows the standardisation of the evaluation procedures
Line 156 allows the identification of both
Line 162 In detail
Line 163 indeed prior selection manual training is needed.
Line 164 beginning the operator must choose samples relating to the
Line 168 and sets up
Better identifies
Line 170 the programme can be saved
Line 169 Are you sure n. of pixels makes sense? What is n?
Line 172-173 max in full?
Line 173 Selection amount refers to cereals
Line 218 The VIS absorptions yielding
Line 219 the Cucurachi
Line 224 was set at 80
Line 228 a solution (remove was prepared before a solution)
Line 229 1 g of
Line 237 Place The phenolic compounds in a new paragraph
Line 177 With respect to the general settings, the Reject choice was set as follows, ie
Line 179 in front and behind, were activated
Line 180 On this screen it was possible to set the speed
Line 184 the solenoid valve remained
Line 188 valves need to the opened at once was set
Line 190 re-write lines 190 to 194
Line 195 -198 re-write could not follow
Figure 2 and text: Why does it refer to 2 choices when there are in fact 3?
Clarify
Line 204 Delete on the other hand as far as milling was concerned.
Instead pls use : The mill used a stainless-steel Alfa
Line 206 Kneading took place for 40
Line 208 used a two-phase
Line 271 Are you sure the word is leave – do you refer to leaf or leaves?
Line 276 A statistical ordination approach was
Line 276 not clear
Line 278 olives. A principal component analysis (PCA) was conducted using the software
Table 2 preferably do not split the word olives
Line 301 obtained from the olives at different
Line 302 green olives were called EVOOmix
Line 303 olives were identified
Line 311 if you refer to a range should you not show highest and lowest figures?
Line 312 same argument as above
Line 317 400 and 500 nm, that correlated to the concentrations of carotenoids and between 500 and 700 nm that corresponded to
Line 323 Did not follow , possibly re-write
Line 333 for the oils obtained from more ripe olives, EVOOSC showed a lower
Line 348 the lowest concentration
Line 349 where it was demonstrated
Line 355 The study showed
Line 357 error rate depended
Line 362 Therefore, the study
Line 370 After the application of the pattern recognition technique, an unsupervised Principal Component Technique (PCA) was applied to the……
Line 372 A biplot of the final PCA was then obtained.
Figure 4 legend Biplot of the PCA conducted on some of the chemical
Line 380 Along the first components
Line 382 – 385 please clarify sentence
Line 390 and optoelectronic ones in particular represent important tools implemented
Line 393 of the assessments, processing speed, capacity to work, (potentially) without interruption etc.
Line 396 to eliminate defective olives
Line 397 the potential of these technologies for use in EVOO production.
Line 398 The oil analyses obtained showed significant differences both chemically
Line 401 while still being able
Line 403 when both unripened and completely ripe
Line 407 thus are commonly concurrently harvested
Line 408 A selection potentially gives the opportunity
Line 411 Future work will test and evaluate
Author Response
Reviewer #3
Line 36 About 82 %
As suggested, we modified the text.
Is reference 3 correct as I could not find reference to % of production in Spain and Italy. If not please correct.
The reference 3 is correct.
Line 41 you refer to mechanical methods yet do not list a mechanical method
Reference is made to mechanical methods because the processes for obtaining EVOO (mentioned in line 41) are mechanical processes (washing, decanting, centrifugation, and filtration).
Line 42 For an EVOO to be defined as such it
As suggested, we modified the text.
Line 45 and the other containing the
As suggested, we modified the text.
Reference 6 - are you sure this is correct. Please go through the first part of the reference section and make sure that the references do indeed say what you quote.
As suggested, we rechecked the references. They are all corrects.
Line 60 in the phase between the green and
As suggested, we modified the text.
Line 62 may vary according to the production region and cultivars milled
As suggested, we modified the text.
Line 63 the genuiness of the
As suggested, we modified the text.
Line 71 allow the determination of the quality of the oil with respect to sensory characteristics
As suggested, we modified the text.
Please check with editor if Food uses US or British spelling. If British then correct colour, flavour
Thank you for your comment. The journal guidelines state that either English or UK English can be used as long as there is consistency. Indeed, we replaced “color” with “colour” and “flavor” with “flavours” in all text.
Line 80 ingredients with a strong fruit flavour.
As suggested, we modified the text.
Line 93 oil, panel testing represents a crucial
As suggested, we modified the text.
Line 93 This allows the standardisation of the evaluation procedures
As suggested, we modified the text.
Line 156 allows the identification of both
As suggested, we modified the text.
Line 162 In detail
As suggested, we modified the text.
Line 163 indeed prior selection manual training is needed.
As suggested, we modified the text.
Line 164 beginning the operator must choose samples relating to the
As suggested, we modified the text.
Line 168 and sets up
As suggested, we modified the text.
Better identifies
As suggested, we modified the text.
Line 170 the programme can be saved
As suggested, we modified the text.
Line 169 Are you sure n. of pixels makes sense? What is n?
We replaced “n.” with “number of pixels”. In this way it makes sense.
Line 172-173 max in full?
Yes, max in full.
Line 173 Selection amount refers to cereals
As suggested, we modified the text.
Line 218 The VIS absorptions yielding
As suggested, we modified the text.
Line 219 the Cucurachi
As suggested, we modified the text.
Line 224 was set at 80
As suggested, we modified the text.
Line 228 a solution (remove was prepared before a solution)
As suggested, we modified the text.
Line 229 1 g of
As suggested, we modified the text.
Line 237 Place The phenolic compounds in a new paragraph
As suggested, we modified the text.
Line 177 With respect to the general settings, the Reject choice was set as follows, ie
As suggested, we modified the text.
Line 179 in front and behind, were activated
As suggested, we modified the text.
Line 180 On this screen it was possible to set the speed
As suggested, we modified the text.
Line 184 the solenoid valve remained
As suggested, we modified the text.
Line 188 valves need to the opened at once was set
As suggested, we modified the text.
Line 190 re-write lines 190 to 194
As suggested, we re-write lines 190 to 194.
Line 195 -198 re-write could not follow
Lines 195-198 mention the classes of olives considered, so it does not need to be re-written.
Figure 2 and text: Why does it refer to 2 choices when there are in fact 3? Clarify
The text and Figure 2 are consistent because 3 classes are mentioned in the text as well as in the figure. In fact, the text states this: “Two classes of olives were identified on the basis of ripeness (“green”, those with an early 196 stage of maturity, and “black”, those that are ripe). In addition, we also considered invaiate olives in the “mixed” class”.
Line 204 Delete on the other hand as far as milling was concerned.
As suggested, we modified the text.
Instead pls use : The mill used a stainless-steel Alfa
As suggested, we modified the text.
Line 206 Kneading took place for 40
As suggested, we modified the text.
Line 208 used a two-phase
As suggested, we modified the text.
Line 271 Are you sure the word is leave – do you refer to leaf or leaves?
As suggested, we replaced with “leaves”.
Line 276 A statistical ordination approach was
As suggested, we modified the text.
Line 276 not clear
As suggested, we re-written a sentence.
Line 278 olives. A principal component analysis (PCA) was conducted using the software
As suggested, we modified the text.
Table 2 preferably do not split the word olives
As suggested, we modified the text.
Line 301 obtained from the olives at different
As suggested, we modified the text.
Line 302 green olives were called EVOOmix
As suggested, we modified the text.
Line 303 olives were identified
As suggested, we modified the text.
Line 311 if you refer to a range should you not show highest and lowest figures?
Line 312 same argument as above
Ranges extreme values were highlighted in the table as suggested.
Line 317 400 and 500 nm, that correlated to the concentrations of carotenoids and between 500 and 700 nm that corresponded to
As suggested, we modified the text.
Line 323 Did not follow, possibly re-write
We replaced a word in order to make the meaning of the sentence clear.
Line 333 for the oils obtained from more ripe olives, EVOOSC showed a lower
As suggested, we modified the text.
Line 348 the lowest concentration
As suggested, we modified the text.
Line 349 where it was demonstrated
As suggested, we modified the text.
Line 355 The study showed
As suggested, we modified the text.
Line 357 error rate depended
As suggested, we modified the text.
Line 362 Therefore, the study
As suggested, we modified the text.
Line 370 After the application of the pattern recognition technique, an unsupervised Principal Component Technique (PCA) was applied to the……
As suggested, we modified the text.
Line 372 A biplot of the final PCA was then obtained.
As suggested, we modified the text.
Figure 4 legend Biplot of the PCA conducted on some of the chemical
As suggested, we modified the text.
Line 380 Along the first components
As suggested, we modified the text.
Line 382 – 385 please clarify sentence PESCARA
That was done, thanks.
Line 390 and optoelectronic ones in particular represent important tools implemented
As suggested, we modified the text.
Line 393 of the assessments, processing speed, capacity to work, (potentially) without interruption etc.
As suggested, we modified the text.
Line 396 to eliminate defective olives
As suggested, we modified the text.
Line 397 the potential of these technologies for use in EVOO production.
As suggested, we modified the text.
Line 398 The oil analyses obtained showed significant differences both chemically
As suggested, we modified the text.
Line 401 while still being able
As suggested, we modified the text.
Line 403 when both unripened and completely ripe
As suggested, we modified the text.
Line 407 thus are commonly concurrently harvested
As suggested, we modified the text.
Line 408 A selection potentially gives the opportunity
As suggested, we modified the text.
Line 411 Future work will test and evaluate
As suggested, we modified the text.
Round 2
Reviewer 1 Report
The authors have incorporated the required changes to the paper.
The paper is much clearer now.
