Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessArticle

Peer-Review Record

MPG-YOLO: Enoki Mushroom Precision Grasping with Segmentation and Pulse Mapping

Agronomy 2025, 15(2), 432; https://doi.org/10.3390/agronomy15020432

by Limin Xie^1,2, Jun Jing¹, Haoyu Wu³, Qinguan Kang⁴, Yiwei Zhao¹ and Dapeng Ye^1,2,*

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Reviewer 3: Anonymous

Agronomy 2025, 15(2), 432; https://doi.org/10.3390/agronomy15020432

Submission received: 9 January 2025 / Revised: 5 February 2025 / Accepted: 7 February 2025 / Published: 10 February 2025

(This article belongs to the Section Soil and Plant Nutrition)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript addresses a timely topic, please find below some suggestions:

In general please check typing errors like "seg-mentation" line 3 or "mush-room" line 21

Title: it is not easy to understand the acronym MPG, I would try to simplify the title

abstract: please add the latin name of the enoki. I would add the main results on the final trial cited at the end.

Keywords: please check the keywords, should not repeat the words of the title.

Introduction: please add the latin name of the Enoki, only the first time you mention it. Please justify better the importance of this crop with specific data in the text. I would avoid to insert Figure 1 at the end of the introduction. Please add at the end of the introduction a clear aim of your research.

Methods: I would call this section "Materials and Methods". I realized from the abstract that a final trial was performed testing in a real situation this new method for processing the mushrooms. Maybe I am mistaken but in case there is no explenation of the metholology of the final tests in real situation.

Results and discussion: Please call this section "Results" as there is a separated section called "Discussion". I would kindly ask to the authors to clarify this aspect, I didn't understand if you tested your vision sistem to improve a real plant for the processing of the mushrooms or if you just tested if your vision system was able the detect the mushrooms.

Discussion: In my opinion in this section it is not very clear the innovation and/or the differences from this study to the literature. Moreover I would focus more on the explanation fo the results. The last part, on future research and limitations, in my opinion should be moved to the conclusions section.

Conclusions: Please see the comments above. I would add some practical benefits the farmers might gain from your research.

Author Response

Dear reviewer, due to the large amount of reply content, I also uploaded my reply letter in the form of attachment, thank you for your understanding

# Reviewer 1

Dear reviewer, I have marked the corresponding changes in the paper in yellow

Comments 1: In general please check typing errors like "seg-mentation" line 3 or "mush-room" line 21

Response 1: Thank you for pointing this out. I agree with this comment. Therefore, I have checked this typo in the full text and corrected it.

Comments 2: Title: it is not easy to understand the acronym MPG, I would try to simplify the title

Response 2: Thank you for pointing this out. I agree with this comment. Therefore, I have changed the title to ‘’MPG-YOLO: Enoki Mushroom Precision Grasping with Segmentation and Pulse-Mapped‘’.MPG means “Mushroom”, “Precision”, “Grasping”

Comments 3: abstract: please add the latin name of the enoki.

Response 3: Thank you for pointing this out. I agree with this comment. Therefore, I have included enoki's Latin, Flammulina velutipes , where it first appears in the Abstract.

Comments 4: abstract: I would add the main results on the final trial cited at the end.

Response 4: Thank you for pointing this out. I agree with this comment. Therefore, I have added the final trial results at the end.

“The constructed mapping relationship achieved a differentiated end-effector drive grasping success rate of 96% and a qualified cutting surface rate of 98%. This effective-ly demonstrates the feasibility of this study and lays the technical foundation for the intelligent upgrade of enoki mushroom cutting equipment.”

I also added relevant information in the article page 22

“A total of 100 enoki mushrooms were tested for root cutting, with 96 successfully achieving effective and accurate differentiated control, resulting in a success rate of 96%. Among these 96 mushrooms, 94 met the quality requirements after root cutting, achieving a qualified cut surface rate of 98%. The results are shown in Figure 20. Surface 1 and 2 represent the root section detached after cutting and the cross-section of the finished enoki mushroom, respectively. MPG surface were complete, with good flat-ness, and the angle between the cut surface and the horizontal plane did not exceed 5°.”

Comments 5: Keywords: please check the keywords, should not repeat the words of the title.

Response 5: Thank you for pointing this out. I agree with this comment. Therefore, I have checked all the keywords to make sure they don't duplicate the title. “machine vision, serial communication, YOLO v8, multi-target recognition, Flammulina velutipes, servo control”

Comments 6: Introduction: please add the latin name of the Enoki, only the first time you mention it.

Response 6: Thank you for pointing this out. I agree with this comment. Therefore, I have included enoki's Latin, Flammulina velutipes, where it first appears in the Introduction.

Comments 7: Introduction: Please justify better the importance of this crop with specific data in the text.

Response 7: Thank you for pointing this out. I agree with this comment. Therefore, I have added specific numbers in the paper to highlight the importance of this crop.

“In 2023, their production exceeded 2.1 million tons, with an export value surpassing 250 million RMB”

Comments 8: Introduction: I would avoid to insert Figure 1 at the end of the introduction.

Response 8: Thank you for pointing this out. I agree with this comment. Therefore, I have moved Figure 1 to Chapter 2 Materials and Methods in the article, page4.

Comments 9: Introduction: Please add at the end of the introduction a clear aim of your research.

Response 9: Thank you for pointing this out. I agree with this comment. Therefore, I have set out the research aim in detail at the end of the introduction.

“In summary, this study addresses the issue of poor cut surface quality in enoki mushrooms, which arises from improper pre-cutting grasping behavior, taking into account the mushrooms' texture, shape, and tenderness. Building on existing research methods, the main objective of this study is to propose a visual detection model, MPG-YOLO v8, specifically designed for enoki mushrooms, to use its output data to control the clamp's actions. Specifically, the approach integrates techniques such as Star Net, SPPECAN, and C2fDStar, which improve both detection speed and accuracy. The model employs masking techniques to address distortion in the grasping region localization caused by anchor boxes. Additionally, through optimizing mask details with mask affiliation judgment, fusion optimization, and target point refinement, the model enhances measurement precision and reduces the likelihood of grasp position prediction errors. Furthermore, by incorporating mapping with PWM control, the model overcomes the limitation of current grasping systems, enabling differentiated grasping actions on enoki mushrooms, and ultimately improving the quality of root cutting.”

Comments 10: Methods: I would call this section "Materials and Methods".

Response 10: Thank you for pointing this out. I agree with this comment. Therefore, I have changed the chapter title to Materials and Methods.

Comments 11: I realized from the abstract that a final trial was performed testing in a real situation this new method for processing the mushrooms. Maybe I am mistaken but in case there is no explenation of the metholology of the final tests in real situation.

Response 11: Thank you for pointing this out. I agree with this comment. Therefore, I have explained in the abstract and conclusion of this paper that the final experiment is carried out in the actual environment.

Abstract: “Experiments in real situation show that”;

Conclusion:” Finally, the method was tested in a factory environment”

Comments 12: Results and discussion: Please call this section "Results" as there is a separated section called "Discussion".

Response 12: Thank you for pointing this out. I agree with this comment. Therefore, I have changed the Section to ‘Results’.

Comments 13: I would kindly ask to the authors to clarify this aspect, I didn't understand if you tested your vision sistem to improve a real plant for the processing of the mushrooms or if you just tested if your vision system was able the detect the mushrooms.

Response 13: Thank you for pointing this out. I agree with this comment. I am sorry that I did not express it accurately in the article, but my research is actually to test visual recognition and drive the claw to perform actions through recognition information. Therefore, I have reorganized the expression in the abstract, introduction and conclusion.

Comments 14: Discussion: In my opinion in this section it is not very clear the innovation and/or the differences from this study to the literature. Moreover I would focus more on the explanation fo the results. The last part, on future research and limitations, in my opinion should be moved to the conclusions section.

Response 14: Thank you for pointing this out. I agree with this comment. Therefore, I have sorted out the discussion part again, and analyzed and discussed the results and the shortcomings and problems in the results.

Discussion

“In order to reduce the labor intensity of manual tasks, most current research on crop detection focuses on the classification or recognition of specific targets. For exam-ple, Li [7] proposed the Cotton-YOLO model for foreign fibers in cotton seed detection; Genno [6] introduced a detection model for apple counting and growth status classifi-cation; Wang [15] presented a detection model for shiitake mushroom dryness classi-fication; and Charisis [19] developed a segmentation model for oyster mushrooms. In contrast, this study not only emphasizes model efficiency and accuracy but also ad-dresses significant detection quality issues, such as mask boundary overflow and in-ternal disconnections. Unlike previous models that focus solely on performance met-rics, the improvements introduced in MPG-YOLO, highlighted in Table 1 and Table 2, target both precision and post-processing optimization. MPG-YOLO v5, for instance, shows a notable 2.68% increase in mAP50:95 for enoki mushroom mask detection compared to YOLO v5, while also reducing the model size by 5MB. This improvement is consistent with the enhancements observed in MPG-YOLO v8 over YOLO v8. This demonstrates the general applicability of the MPG improvement method across the YOLO series, offering an outstanding combination of speed and accuracy. Furthermore, the integration of a mask optimization algorithm in the post-processing stage, as shown in Figure 17 and Table 4, significantly enhances the visualization quality of the enoki mushroom grabbing area mask. This innovation emphasizes the strength of this study in improving the visual output of the detection, an area often overlooked in previous research. While research in crop detection quality has also focused on phenotypic measurements—such as Kim [9] visual system for corn stalk height measurement, Lu [16] algorithm for mushroom cap diameter, and others—the majority of these methods treat measurement as an isolated task. In contrast, this study uniquely synchronizes localization and measurement by encapsulating both functions together, resulting in superior real-time performance. Notably, Zhou [35] proposed a grape cluster picking point localization method similar to the approach in this study; however, these meas-urement methods did not use the recognition results as driving data for further appli-cations. In contrast, this study not only focuses on improving the model's quantifica-tion metrics but also enhances detection quality and uses the detection results as driv-ing data to control the lower machine. Although Zhao [36] developed a detection algo-rithm for driving the black fungus harvesting robot, regrettably, no attention was paid to further improving the detection quality. It is undeniable that this study also has some limitations. Although the mask merging algorithm can effectively realize the parent-child determination and merging, a few low-quality masks still exist (shown in Figure 17). Figures.17(5b) and 17(6b) show that the optimized ROI still contains noticeable internal gaps. Several factors contribute to this issue: 1. The quality of the initial mask is poor. The regions 4a, 5a, and 6a exhibit varying degrees of internal gaps, with 5a and 6a being more severely affected. Consequently, the mask quality after processing 5b and 6b does not improve to the same level as that of 4b. 2. Occlusion and overlap issues arise. It is evident that the ROI region of the enoki mushrooms in 17(5a) and (6a), particularly 6a, overlaps with the enoki mushroom bodies of other clusters, leading to occlusion. This overlap intro-duces noise, adversely impacting feature extraction and reducing the mask quality. An analysis of the locations where low-quality masks occur reveals that the most severe internal disconnections are found along the boundaries of enoki mushrooms. This re-gion is also where overlapping enoki mushrooms are most likely to cause recognition confusion. Figure 17(1a-3a) and 17(7a-8a) do not exhibit this disconnection issue because there is no overlap in these cases. Therefore, it can be inferred that the primary cause of internal disconnections in the masks is the model’s insufficient ability to recognize and discern boundary features in cases of enoki mushroom overlap. This issue can be addressed by enhancing the images to increase data diversity. Additionally, Figure 19 illustrates a noticeable difference in the detection accuracy of the ROI before and after integrating the models. Before integration, ignoring the ROIs with incom-plete shots, it is obvious that there are ROIs with complete shots that can be artificially recognized as ROIs that cannot be successfully detected. However, applying a model ensemble approach successfully detected these previously unrecognized regions. It suggests that the robustness of individual sub-models still requires enhancement.”

Comments 15: Conclusions: Please see the comments above. I would add some practical benefits the farmers might gain from your research.

Response 15: Thank you for pointing this out. I agree with this comment. Therefore, I have added the economic benefits that this research may bring to the enterprise at the end of the conclusion.

“The method is highly significant for advancing the automation of enoki mushroom production equipment and improving business profitability.”

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The subject of the research is intriguing, yet it needs considerable improvement. The authors should work on enhancing particular areas to ensure greater clarity and depth.
- In the abstract, check the English “ac-curacy” and lines (34-37), as the sentence is quite lengthy.
- In keywords, change “You Only Look Once” to YOLOv8.
- Mention the key findings in the abstract section to highlight the importance of the suggested approach.
- In the introduction, the study's limitations should be addressed, along with an explanation of how the proposed approach can resolve these issues.
- Highlight the main and sub-objectives, as they are not clearly outlined by the end of the introduction.
- The resolution of Figure 1 needs to be improved.
- Line 119, Section 2: The heading should be written as 'Materials and Methods' instead of 'Methods.'
- Line (143), “The results in section 2.2.1 provided a region reference for manual labeling”. Rewrite with deleting “The results in section 2.2.1”.
- Line (166, 158), no need to repeat the information; stating it once is enough.
- Lines (259-272), please replace these steps with pseudo-code.
- Lines (318-319), it would be better to separate the equations from the text.
- (Lines 380-386), It would be better to create a new section called "Software for data analysis.
- Lines (387-397), the section titled "Model Evaluation Metrics" is fine as it is.
- Line (400), rewrite ““①,” “②,” and “③””, describe in a professional manner.
- Figure 13 needs improvement. The details will become clearer when zooming in.
- Line (398), in the 'Results and Discussion' section, change it to 'Results' since there is a separate section called 'Discussion.'
This paper requires significant revisions before it can be considered complete. I would recommend a major revision. Thank you.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

Dear reviewer, since the reply contains a lot of content and pictures, I also uploaded the reply letter as an attachment. Thank you for your understanding

# Reviewer 2

Dear reviewer, I have marked the corresponding changes in the paper in green

Comments 1: - In the abstract, check the English “accuracy” and lines (34-37), as the sentence is quite lengthy.

Response 1: Thank you for pointing this out. I agree with this comment. Therefore, I have checked this and corrected it.

Comments 2: - In keywords, change “You Only Look Once” to YOLOv8.

Response 2: Thank you for pointing this out. I agree with this comment. Therefore, I have changed You Only Look Once to YOLOv8.

Comments 3: - Mention the key findings in the abstract section to highlight the importance of the suggested approach.

Response 3: Thank you for pointing this out. I agree with this comment. Therefore, I have re-emphasized the important results of this study in the abstract.

Abstract

“The flatness of the cut surface of enoki mushrooms (Flammulina velutipes) directly affects their quality classification. In current enoki mushroom automatic cutting equip-ment, mushrooms are grasped by a clamping mechanism and delivered to the cutting end. However, the traditional mechanical clamping structure cannot adjust the grasping action based on the size of individual mushrooms, leading to significant deformation before and after cutting, thereby reducing product quality. To address this issue, this study proposes an improved method that uses visual results to control the behavior of the execution end, ensuring cut quality. The method incorporates Star Net, SPPECAN (reconstructed SPPF with efficient channel attention), and C2fDStar (C2f with Star Net and deformable convolution) to enhance YOLOv8n-seg for more efficient feature ex-traction and processing, while reducing computational complexity and feature loss. Additionally, a mask ownership judgment and merging optimization algorithm is in-troduced to address issues such as positional offset, internal disconnection, and boundary instability when predicting mushroom grasping areas. Based on this, an op-timization method for grasping parameters is proposed, measuring the optimal region width with the optimized centroid as the reference point. Finally, a region width-to-PWM mapping model is constructed to achieve rapid conversion from camera measurement to gripper control, solving the problem of traditional mechanical end ef-fectors' inability to adaptively adjust the grasping action. Experiments in real situation show that the mean average precision (mAP50:95) for grasping area mask reached 0.743, an increase of 4.5% compared to YOLOv8, with an average detection speed of 10.3 ms and a target width measurement error of only 0.14%. The constructed mapping rela-tionship achieved a differentiated end-effector drive grasping success rate of 96% and a qualified cutting surface rate of 98%. This effectively demonstrates the feasibility of this study and lays the technical foundation for the intelligent upgrade of enoki mushroom cutting equipment.”

Comments 4: - In the introduction, the study's limitations should be addressed, along with an explanation of how the proposed approach can resolve these issues.

Response 4: Thank you for pointing this out. I agree with this comment. Therefore, I have illustrated in the end of the introduction the contribution made by the proposed method in this paper to solve these problems

Comments 5: - Highlight the main and sub-objectives, as they are not clearly outlined by the end of the introduction.

Response 5: Thank you for pointing this out. I agree with this comment. Therefore, I have added this to the end of the introduction

Comments 6: - The resolution of Figure 1 needs to be improved.

Response 6: Thank you for pointing this out. I agree with this comment. Therefore, I have improved the resolution of figure 1.

Comments 7: - Line 119, Section 2: The heading should be written as 'Materials and Methods' instead of 'Methods.'

Response 7: Thank you for pointing this out. I agree with this comment. Therefore, I have changed Section2 to Materials and Methods.

Comments 8: - Line (143), “The results in section 2.2.1 provided a region reference for manual labeling”. Rewrite with deleting “The results in section 2.2.1”.

Response 8: Thank you for pointing this out. I agree with this comment. Therefore, I have rewritten it in new section 2.3.2.

“Reference for manual labeling was built.”

Comments 9: - Line (166, 158), no need to repeat the information; stating it once is enough.

Response 9: Thank you for pointing this out. I agree with this comment. Therefore, I have cut out the first repetition.

“The image content was labeled using Labelme’s polygonal pattern: the enoki mush-room was labeled “enoki,” and the grabbing region was labeled “pick region.” We augmented the data using Gaussian blurring, mirror flipping, and brightness ad-justments to simulate poor vision.”

Comments 10: - Lines (259-272), please replace these steps with pseudo-code.

Response 10: Thank you for pointing this out. I agree with this comment. Therefore, I have replaced it with pseudo-code in page 10.

BEGIN

enoki_tags = set of all "enoki" tags

pick_region_tags = set of all "pick_region" tags

group = empty set

minimum_distance_for_pair = infinity

FOR each enoki IN enoki_tags DO

FOR each pick_region IN pick_region_tags DO

enoki_contour = read contour pixels of enoki

pick_region_contour = read contour pixels of pick_region

enoki_center_of_mass = calculate centroid of enoki_contour

pick_region_center_of_mass = calculate centroid of pick_region_contour

IF enoki_center_of_mass and pick_region_center_of_mass meet regional distribution conditions THEN

distance = calculate Euclidean distance between enoki_center_of_mass and pick_region_center_of_mass

IF distance < minimum_distance_for_pair THEN

minimum_distance_for_pair = distance

group.add((enoki, pick_region))

END IF

END FOR

filtered_group = remove duplicate label pairs from group

RETURN filtered_group

END

Comments 11: - Lines (318-319), it would be better to separate the equations from the text.

Response 11: Thank you for pointing this out. I agree with this comment. Therefore, I have separated the equations from the text in page 12.

Comments 12: - (Lines 380-386), It would be better to create a new section called "Software for data analysis.

Response 12: Thank you for pointing this out. I agree with this comment. Therefore, I have created a new section named Software and Hardware for data analysis in page 14.

Comments 13: - Lines (387-397), the section titled "Model Evaluation Metrics" is fine as it is.

Response 13: Thank you very much.

Comments 14: - Line (400), rewrite ““①,” “②,” and “③””, describe in a professional manner.

Response 14: Thank you for pointing this out. I agree with this comment. Therefore, I have rewritten it as “Region A,” “Region B,” and “Region C”. I also modified the corresponding content in Figure 13.

(a) (b)

Comments 15: - Figure 13 needs improvement. The details will become clearer when zooming in.

Response 15: Thank you for pointing this out. I agree with this comment. Therefore, I have improved the Figure 14 (it is Figure 14 now).

Comments 16: - Line (398), in the 'Results and Discussion' section, change it to 'Results' since there is a separate section called 'Discussion.'

Response 16: Thank you for pointing this out. I agree with this comment. Therefore, I have changed 'Results and Discussion' section to 'Results'.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

- The English of the manuscript should be improved.

- The objective of the study is not well stated in the Abstract.

- Overall, the Abstract is not well structured and lack coherence between its paragraphs.

-Also, the summarized methodology of carrying out this work is lacking from the Abstract.

- Authors should pay attention to writing some words such as segmentation, accuracy, etc.

- Figure 1 can be more appreciable in Material and Methods. We suggest to remove it from the Introduction.

- For Material and Methods, we suggest that authors add a first sub-section entitled "Plant material" with relevant data about enoki mushroom used in the study.

- The obtained results in this paper are relevant but not sufficiently discussed with similar works. Thus, it was pointed out that 50% of used references were reported in the Introduction. The other 50% of references were reported in Material and Methods. Only, 4 references were cited in the Discussion section that we can qualify as "not suitable".

- The conclusion is too long and should be summarized. In addition, some words should be avoided (e.g. takeaways, ablation, etc.).

- Overall, we suggest to summarize this paper with only relevant information especially in Material and Methods section. In actual form, it seems to be like a scientific report.

Author Response

Dear reviewer, since the reply contains a lot of contents, I also uploaded the reply letter as an attachment. Thank you for your understanding

# Reviewer 3

Dear reviewer, I have marked the corresponding changes in the paper in blue

Comments 1: - The English of the manuscript should be improved.

Response 1: Thank you for pointing this out. I agree with this comment. Therefore, I have tried to improve the English in the manuscript.

Comments 2: - The objective of the study is not well stated in the Abstract.

Comments 3: - Overall, the Abstract is not well structured and lack coherence between its paragraphs.

Comments 4: -Also, the summarized methodology of carrying out this work is lacking from the Abstract.

Response 2-4: Thank you for pointing this out. I agree with this comment. Therefore, I have rewritten my abstract and sorted it out according to four parts: research objectives, research methods, results and significance.

Abstract

Comments 5: - Authors should pay attention to writing some words such as segmentation, accuracy, etc.

Response 5: Thank you for pointing this out. I agree with this comment. Therefore, I have checked and corrected the text for irregular spelling or wrong words.

Comments 6: - Figure 1 can be more appreciable in Material and Methods. We suggest to remove it from the Introduction.

Response 6: Thank you for pointing this out. I agree with this comment. Therefore, I have removed Figure 1 to Section2 Material and Methods.

Comments 7: - For Material and Methods, we suggest that authors add a first sub-section entitled "Plant material" with relevant data about enoki mushroom used in the study.

Response 7: Thank you for pointing this out. I agree with this comment. Therefore, I have added a section named Plant material in page 4, section 2.1, which systematically introduces the specifications of enoki mushrooms used in this study.

Comments 8: - The obtained results in this paper are relevant but not sufficiently discussed with similar works. Thus, it was pointed out that 50% of used references were reported in the Introduction. The other 50% of references were reported in Material and Methods. Only, 4 references were cited in the Discussion section that we can qualify as "not suitable".

Response 8: Thank you for pointing this out. I agree with this comment. Therefore, I have rewritten the discussion part and used 7 references respectively to reflect the differences and innovations of this study from two perspectives: identification orientation research direction and phenotypic measurement research direction.

While research in crop detection quality has also focused on phenotypic meas-urements—such as Kim [9] visual system for corn stalk height measurement, Lu [16] algorithm for mushroom cap diameter, and others—the majority of these methods treat measurement as an isolated task. In contrast, this study uniquely synchronizes localization and measurement by encapsulating both functions together, resulting in superior real-time performance. Notably, Zhou [35] proposed a grape cluster picking point localization method similar to the approach in this study; however, these meas-urement methods did not use the recognition results as driving data for further appli-cations. In contrast, this study not only focuses on improving the model's quantifica-tion metrics but also enhances detection quality and uses the detection results as driv-ing data to control the lower machine. Although Zhao [36] developed a detection algo-rithm for driving the black fungus harvesting robot, regrettably, no attention was paid to further improving the detection quality.”

Comments 9: - The conclusion is too long and should be summarized. In addition, some words should be avoided (e.g. takeaways, ablation, etc.).

Response 9: Thank you for pointing this out. I agree with this comment. Therefore, I have condensed the conclusion section to keep only the most important data results and to avoid the use of inappropriate words.

Comments 10: - Overall, we suggest to summarize this paper with only relevant information especially in Material and Methods section. In actual form, it seems to be like a scientific report.

Response 10: Thank you for pointing this out. I agree with this comment. However, I believe that the detailed description in the Materials and Methods section is crucial for fully presenting the experimental design and data processing procedures. Therefore, we have decided to retain these details to ensure the transparency and reproducibility of the research. I hope you will forgive me, thank you.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors, thank you for addressing my comments and congratulations for your excellent work.

I would kindly ask just a small improvement: please write the latin name of the mushroom in italics together with the abbreviation of the name of the classifier.

Moreover are you sure it isn't "Flammulina filiformis"? Please check.

Good luck for the publication of your paper.

Author Response

Comments 1:Dear Authors, thank you for addressing my comments and congratulations for your excellent work.

I would kindly ask just a small improvement: please write the latin name of the mushroom in italics together with the abbreviation of the name of the classifier.

Moreover are you sure it isn't "Flammulina filiformis"? Please check.

Good luck for the publication of your paper.

Response 1: Thank you for pointing this out. I agree with this comment. Therefore, I changed it to "Flammulina filiformis" as required by the font, along with the classifier's name. I'm sorry for my mistake

Flammulina filiformis Z.W. Ge, X.B. Liu & Zhu L. Yang

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript was significantly improved following the modifications. Thank you.

Author Response

Thank you! It is an honor for my article to receive your recognition!

Reviewer 3 Report

Comments and Suggestions for Authors

The revised version of this manuscript is well written and addressed generally our comments. However, this paper still needs some refinements:

- The English style of this paper still need an improvement.

- L21-22: "to control the behaviour" is not suitable. Please change with appropriate words...

- The abstract is still not well written. Authors dedicated more importance to justifying results instead of stating clear objective and results.

- Pease italic style for 'Flammulina velutipes'.

- Figure 2 captions are given in Chinese !

- The pseudocode (L299-L329) is not appropriate. It can be removed.

Comments on the Quality of English Language

I advice to check the English style of this paper by a native speaker or proceed with English editing ...

Author Response

Dear Reviewer, due to the inclusion of images in the reply, I have also uploaded the reply file for your review. The relevant sections in the manuscript have been marked in blue.

The revised version of this manuscript is well written and addressed generally our comments. However, this paper still needs some refinements:

Comments 1:- The English style of this paper still need an improvement.

Response 1: Thank you for pointing this out. I agree that the English style of the paper could be improved. As a result, I have revised sections such as the abstract, conclusion, and discussion to enhance clarity, conciseness, and overall readability.

Comments 2:- L21-22: "to control the behaviour" is not suitable. Please change with appropriate words...

Response 2: Thank you for pointing this out. I agree with this comment. But the phrase has been removed because of changes to the abstract.

Comments 3:- The abstract is still not well written. Authors dedicated more importance to justifying results instead of stating clear objective and results.

Response 3: Thank you for pointing this out. I agree with this comment. Therefore, I have made adjustments to the abstract to make the research objectives and results clearer. This study aims to address the deformation issues in enoki mushroom cutting caused by conventional automatic equipment’s inability to adjust grasping movements based on mushroom size. I hope this revision meets your approval.

Abstract

The flatness of the cut surface in enoki mushrooms (Flammulina filiformis Z.W. Ge, X.B. Liu & Zhu L. Yang) is a key factor in quality classification. However, conventional automatic cutting equipment struggles with deformation issues due to its inability to adjust the grasping force based on individual mushroom sizes. To address this, we propose an improved method that integrates visual feedback to dynamically adjust the execution end, enhancing cut precision. Our approach enhances YOLOv8n-seg with Star Net, SPPECAN (a reconstructed SPPF with efficient channel attention), and C2fDStar (C2f with Star Net and deformable convolution) to improve feature extraction while reducing computational complexity and feature loss. Additionally, we introduce a mask ownership judgment and merging optimization algorithm to correct positional offsets, internal disconnections, and boundary instabilities in grasping area predictions. Based on this, we optimize grasping parameters using an improved centroid-based region width measurement and establish a region width-to-PWM mapping model for precise conversion from visual data to gripper control. Experiments in real situation settings demonstrate the effectiveness of our method, achieving a mean average precision (mAP50:95) of 0.743 for grasping area segmentation, a 4.5% improvement over YOLOv8, with an average detection speed of 10.3 ms and a target width measurement error of only 0.14%. The proposed mapping relationship enables adaptive end-effector control, resulting in a 96% grasping success rate and a 98% qualified cutting surface rate. These results confirm the feasibility of our approach and provide a strong technical foundation for the intelligent automation of enoki mushroom cutting systems.

Comments 4:- Pease italic style for 'Flammulina velutipes'.

Response 4: Thank you for pointing this out. I agree with this comment. Therefore, I adjusted it to italic style.

Comments 5:- Figure 2 captions are given in Chinese !

Response 5: Thank you for pointing this out. I'm sorry for my oversight. I have adjusted the text in the picture to English. (in page 4)

Comments 6:- The pseudocode (L299-L329) is not appropriate. It can be removed.

Response 6: Thank you for pointing this out. Regarding the inclusion of pseudocode, it was actually suggested by another reviewer. In order to accommodate both your feedback and the suggestion from the other reviewer, I have revised the original pseudocode to better align with the academic style typically seen in scholarly papers. Thank you once again for your understanding and support. I highly appreciate your comments and hope these revisions meet your expectations.(in page 9)

Algorithm: Generate label pairs based on region distribution and center of mass distance

Input:

enoki_tags: Set of "enoki" tags

pick_region_tags: Set of "pick region" tags

Output:

group: Set of label pairs {enoki_tag, pick_region_tag}

Initialize group ← ∅

For each enoki_tag ∈ enoki_tags do

For each pick_region_tag ∈ pick_region_tags do

// Step 2a: Read contour pixels corresponding to the labeled objects

contour_pixels ← ReadContourPixels(enoki_tag, pick_region_tag)

Calculate center_of_mass(enoki_tag, pick_region_tag) for each contour

Filter pairs of centers of mass that meet region distribution conditions

For each pair of centers of mass (Class A, Class B) do

// Step 5a: Compute Euclidean distance (ED) between centers of mass

ED ← EuclideanDistance(center_of_mass_A, center_of_mass_B)

// Step 5b: Track the minimum Euclidean distance for each pair

min_ED ← min(min_ED, ED)

For each pair {enoki_tag, pick_region_tag} do

// Step 6a: If minimum ED value is obtained, add label pair to group

if ED = min_ED then

group ← group ∪ {enoki_tag, pick_region_tag}

Remove duplicates from group

Return group

Author Response File: Author Response.pdf

Article Menu

MPG-YOLO: Enoki Mushroom Precision Grasping with Segmentation and Pulse Mapping

Further Information

Guidelines

MDPI Initiatives

Follow MDPI