Blackberry Growth Monitoring and Feature Quantification with Unmanned Aerial Vehicle (UAV) Remote Sensing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

The work presented in the Manuscript, entitled ^„ Blackberry Growth Monitoring and Feature Quantification with Unmanned Aerial Vehicle (UAV) Remote Sensing ^„. The manuscript is well written and can be accepted after minor revision.

 

·         What is the novelty of the work?. What is new in your work that makes a difference in the body of knowledge?

 

·         What is the benefit of Figure 13? Correlation matrix of number of plants and vegetation coverage (%) over 11 weeks.

 

·         Please, write the practical applications of your work in a separate section, before the conclusions and provide your good perspectives.

 

·         Please write about the limitations of this work in details in conclusion section.

 

•  The important results should be added in abstract.
• The conclusion of the abstract should be added at the end. Please remove the last sentence in abstract.
• Keywords should be arranged alphabetic.
• Please in introduction; write about the basic of used RGB in this study.
• What is the hypothesis in this study?
• Line 218. Please change 2.4. Statistical Analysis to Data Analysis
• Figure 6. Please could you remove the scatter points such as the last 3 points which in the corner from 6 to 8 to improve R2.
• The resolution of the figures from 9 to 12 should be improved.

 

Author Response

Response to Reviewer 1 Comments

The work presented in the Manuscript, entitled ^„ Blackberry Growth Monitoring and Feature Quantification with Unmanned Aerial Vehicle (UAV) Remote Sensing ^„. The manuscript is well written and can be accepted after minor revision.

Comment 1:  What is the novelty of the work? What is new in your work that makes a difference in the body of knowledge?

Response 1: Thank you for asking about this. The novelty of this work is based on the application of UAV-based remote sensing to accurately measure flower and vegetation coverage in blackberries, a specialty crop that presents unique challenges for traditional phenotyping due to its dense and hedge-like structure. The novelty of using the same Hue, Saturation, Brightness (HSB) parameters developed for object segmentation were consistently applied throughout the 11-week period to segment flowers and vegetation in the blackberry data set collected at the experimental site. This method guaranteed consistency in feature identification throughout the study period while maintaining the distinct characteristics of flowers and vegetation in each image. The novelty of using these HSB values to easily automate a simpler workflow for automation compared to extended object detection for agronomic or breeding purposes.  Additionally, by developing the Flower-Vegetation Ratio (FVR) as a reliable, quantitative metric for flowering intensity and coverage, this research introduces an objective method and offers a scalable alternative to manual counting. Also, basing the flower area on the vegetation area provides a novel approach based on the differences in the number of plants affecting plot sizes. The study’s temporal feature analysis also provides new insights into the flowering and vegetation dynamics of different blackberry genotypes (primocane and floricane), enhancing understanding of their phenological patterns and supporting targeted breeding and management decisions.

This work advances the body of knowledge by establishing a validated, high-throughput methodology that can efficiently capture blackberry flowering traits over large areas, significantly reducing labor and increasing consistency. The use of open-source image analysis tools and the UAV-based approach has created new opportunities for future research in specialty crop management. Additionally, by addressing limitations and proposing paths for methodological improvement, this study sets a foundation for continued innovation in precision agriculture, potentially extending its impact to similar horticultural crops. This research finally makes a significant contribution by demonstrating that UAVs, combined with image analysis, can provide reliable, actionable data for enhancing blackberry cultivation and broader phenotyping applications.

 Comment 2: What is the benefit of Figure 13? Correlation matrix of number of plants and vegetation coverage (%) over 11 weeks.

Response 2: Thank you for your question on Figure 13. We added the correlation matrix of the number of plants and vegetation coverage (%) over 11 weeks to offer valuable insights into the consistency of vegetation coverage over time and the growth dynamics. Vegetation coverage was used as an indicator variable of plant or cane count due to the difficulty to visually estimate number of canes per plant within a plot. The idea was to check the level of agreement between vegetation coverage and plant count.

Comment 3: Please, write the practical applications of your work in a separate section, before the conclusions and provide your good perspectives.

Response 3: Thank you for pointing this out. We have added a separate section providing practical applications of the study between lines 527 and 557.

Comment 4: Please write about the limitations of this work in details in conclusion section.

Response 4: Thank you for pointing this out. We have added detailed limitations of this work in the conclusion section between lines 572 and 58.

Comment 5: The important results should be added in abstract.

Response 5: Thank you for pointing this out. We have added important results to the abstract between lines 29 and 31.

Comment 6: The conclusion of the abstract should be added at the end. Please remove the last sentence in abstract.

Response 6: Thank you for pointing this out. We have added a conclusion of the abstract at the end and removed the last sentence originally in the abstract.

Comment 7: Keywords should be arranged alphabetic.

Response 7: Thank you for pointing this out. We have arranged keywords in alphabetical order in lines 37 and 38

Comment 8: Please in introduction; write about the basic of used RGB in this study.

Response 8: Thank you for your suggestion. We have added basic used RGB in this study between lines 98 and 107

Comment 9: What is the hypothesis in this study?

Response 9: Thank you for pointing this out. This study hypothesized that UAV-based remote sensing, along with image analysis techniques, can effectively quantify blackberry flower and vegetation coverage, offering a reliable, scalable, and objective alternative to conventional, labor-intensive phenotyping methods. This methodology was expected to uncover distinct temporal flowering patterns and growth dynamics, which can be utilized for precision agriculture practices. Vegetation specific flower density quantification could provide efficient assessment for plant breeding data such as flowering time, density, level of injury from early-spring frost events and other relevant phenotypic features.

Comment 10: Line 218. Please change 2.4. Statistical Analysis to Data Analysis

Response 10: Thank you for your suggestion. We have duly made changes to this in lines 240.

Comment 11: Figure 6. Please could you remove the scatter points such as the last 3 points which in the corner from 6 to 8 to improve R2.

Response 11: Thank you for pointing this out. We have removed the scatter points accordingly and provided a new graph and improved R² in Figure 7. The figure number has changed as we added a new figure based on other reviewer’s comment.

Comment 12: The resolution of the figures from 9 to 12 should be improved.

Response 12: Thank you for your suggestion. We have added new figures with high resolution for figures 9 to 12. Now 10 to 13 as we added a new figure based on other reviewer’s comment.

Reviewer 2 Report

Comments and Suggestions for Authors

The aim of this study is to develop a methodology for the accurate assessment of blackberry flowers and vegetation using remote sensing by unmanned aerial vehicles (UAVs). In addition, the study was aimed at establishing the image processing parameters necessary for an autonomous UAV subsystem.

This study presents a new methodology for accurately quantifying the ratio of flowering and vegetable crops in blackberries using (UAVs) equipped with RGB sensors. Thanks to the integration of remote sensing technologies and advanced image processing techniques, the authors were able to evaluate and track the flowering dynamics of various blackberry varieties.

The results of the study show that FVR is a reliable indicator of flowering intensity, which demonstrates a moderately strong correlation with the actual number of flowers obtained using visual estimates. The Pearson correlation value the difference between FVR and visual estimates is relatively large, which explains about 48% of the variance. This connection is crucial, given the difficulties associated with manually counting colors using methods that require a lot of work and are subject to human bias.

The article is well written and well structured, overall, with a clear introduction, methodology, results and the main part. The authors effectively communicate the results of their research, conclusions and their significance. The article is the result of the collaboration of several authors with different experiences, which increases the overall quality and depth of the research.

But, despite the positive results of the study, the article requires some improvements:

1. You need to select a different palette for Fig. 13 for better readability. 2. It is necessary to add the section "Mathematical formulation of the problem". 3. For ease of reading, enter the abbreviation table. 4. Bring the abbreviation to one style. For example, line 430 Flower Vegetation Ratio (FVR) and line 88 unmanned aerial vehicles (UAVs). 5. At the end of Section 1, you need to add a brief structure of the article 6. In the article, you need to add a comparative analysis with the results of other work on your dataset. 7. There is no description of Fig. 6 on page 8.

Author Response

agriengineering-3288845 - Comments

Response to Reviewer 2 Comments

The aim of this study is to develop a methodology for the accurate assessment of blackberry flowers and vegetation using remote sensing by unmanned aerial vehicles (UAVs). In addition, the study was aimed at establishing the image processing parameters necessary for an autonomous UAV subsystem.

This study presents a new methodology for accurately quantifying the ratio of flowering and vegetable crops in blackberries using (UAVs) equipped with RGB sensors. Thanks to the integration of remote sensing technologies and advanced image processing techniques, the authors were able to evaluate and track the flowering dynamics of various blackberry varieties.

The results of the study show that FVR is a reliable indicator of flowering intensity, which demonstrates a moderately strong correlation with the actual number of flowers obtained using visual estimates. The Pearson correlation value the difference between FVR and visual estimates is relatively large, which explains about 48% of the variance. This connection is crucial, given the difficulties associated with manually counting colors using methods that require a lot of work and are subject to human bias.

The article is well written and well structured, overall, with a clear introduction, methodology, results and the main part. The authors effectively communicate the results of their research, conclusions and their significance. The article is the result of the collaboration of several authors with different experiences, which increases the overall quality and depth of the research.

But, despite the positive results of the study, the article requires some improvements:

Comment 1: You need to select a different palette for Fig. 13 for better readability.

Response 1: Thank you for pointing this out. We have provided a different color palette for figure 13 to ensure better readability. Now figure 14 as we added a new figure based on reviewer’s comment.

Comment 2: It is necessary to add the section "Mathematical formulation of the problem".

Response 2: Thank you for your suggestion, this was an important point to state. However, we do not have enough material to add a new section for "Mathematical formulation of the problem". To address this, we have included a paragraph description of the logic that was applied and represents the mathematical formulation of the problem in Section 2.4 between lines 262 and 271 we have also provided a flow-chart in Figure 5 for this.

Comment 3: For ease of reading, enter the abbreviation table.

Response 3: Thank you for your suggestion. We have added an abbreviation table in Appendix in Table A1 for ease of reading.

Comment 4: Bring the abbreviation to one style. For example, line 430 Flower Vegetation Ratio (FVR) and line 88 unmanned aerial vehicles (UAVs).

Response 4: Thank you for pointing this out. We have realigned all abbreviations to one style. 

Comment 5: At the end of Section 1, you need to add a brief structure of the article

Response 5: Thank you for pointing this out. We appreciate your suggestion, we did not include a brief structure of the article in our publications as in addition to the template provided by the AgEngineering journal, we also follow the American Society of Agricultural and Biological Engineering (ASABE) manuscript standards.

Comment 6. In the article, you need to add a comparative analysis with the results of other work on your dataset.

Response 6: Thank you for pointing this out. We have already provided comparative analysis in the discussion section. These can be found between lines 366 and 370. However, we have provided further comparative analysis in the Discussion section between lines 426 and 488.

Comment 7: There is no description of Fig. 6 on page 8.

Response 7: Thank you for pointing this out. We have a description before Figure 6 in lines now Figure 7 between lines 310 and 312.

 

Reviewer 3 Report

Comments and Suggestions for Authors

1. Figure 5, are these RGB images stitched? If the image is stitched, how to do it? If not, how to get the orthophoto?

2. The UAV used in this manuscript supports the RTK system, Is it active?

3. Figure 6, why the flower-vegetation ratio is not continuous? It looks like the interval is 0.1%. Also, the flower estimates area.

4. I found there are some vertical lines in Figure 10, and Figure 12. Why? Is the measurement finished twice a weak?

5. Is the experiment field close to the mountain? Maybe the isopach map will be interesting.

6. The data was only analyzed by some traditional methods, please authors consider open-source data for better utilization of data. Such as some AI methods.

7. Figure 14, is this FF or PF?

8. The supplementary table is not downloadable in the review system. Please check it.

 

 

Author Response

agriengineering-3288845 - Comments

Response to Reviewer 3 Comments

Comment 1: Figure 5, are these RGB images stitched? If the image is stitched, how to do it? If not, how to get the orthophoto?

Response 1: Thank you for pointing this out. We did not stitch these RGB images, these are original RGB images captured by the UAV at 12m above ground level which have been clipped and rotated to determine selected plots. A new flow chart has been (Figure 5) has been added to make this easy to understand. Orthophotos were created from 80% overlap and side-lapped RGB images taken from UAV at 12m altitude. This orthomosaic was represented in Figures 1 and 15. We are going to conduct orthomosaic-driven flower count analysis in our next work. The purpose of using individual images for this research was to create variables for UAV-based instantaneous and on-the-go flower counting subsystem development. We have also added a description of how we created an orthophoto under section 2.3 between lines 228 and 238.

Comment 2: The UAV used in this manuscript supports the RTK system, Is it active?

Response 2: Thank you for pointing this out. The RTK system was not active during data collection. However, the system used had +\- 15cm positioning accuracy.

Comment 3: Figure 6, why the flower-vegetation ratio is not continuous? It looks like the interval is 0.1%. Also, the flower estimates area.

Response 3: Thank you for pointing this out. This figure represents flower-vegetation ratio for the 123 plots over the 11-week period. This has resulted in the majority of data points overlapping and clustered in the lower-section of the graph. Flower-vegetation ratio is not continuous because some plots had more flowers and less vegetation in comparison to others while some plots had no flower but dense vegetation, this uneven distribution of flower and vegetation ratio in the plots could be the potential reason behind the data distribution in Figure 6 (Figure 7 now). Even if blackberry plants produced consistent flower-vegetation ratio, because the images cannot detect flowers that are hidden under leaves the FVR among all the plots in Figure 6 (7) did not show a continuous pattern. 

Comment 4: I found there are some vertical lines in Figure 10, and Figure 12. Why? Is the measurement finished twice a weak?

Response 4: Thank you for pointing this point. The vertical line in Figure 12 (13 now) was due to double cropping of primocane plots (eg. S585). To address this, we have added the letter “B” to the plot and the trendlines per genotype has been removed. Measurements were, however, conducted once a week, every Wednesday at Solar-noon.

Comment 5: Is the experiment field close to the mountain? Maybe the isopach map will be interesting.

Response 5: Thank you for asking. The experimental field is not close to the mountain, the stated region “in the foothills of the Ozark mountains” was to notify the terrain, soil type and climate for the experimental location. Thank you for your suggestion, we believe at this stage that the focus of this article is not solely on mapping techniques. While an isopach map could provide insights into specific soil layers with apparent influence on nutrient availability, water retention and root development, none of our research objectives require elevation data. However, there was approximately 5 percent slope in the experimental site. We have provided detailed description of slope between lines 262 and 271.

Comment 6: The data was only analyzed by some traditional methods, please authors consider open-source data for better utilization of data. Such as some AI methods.

Response 6: Thank you for the suggestion. We appreciate the importance of applying advanced analytical techniques, including AI methods, to enhance data utilization and extract more nuanced insights. While this study primarily focused on validating traditional methods for establishing a baseline, we acknowledge that incorporating open-source AI techniques could offer additional value in future analyses. In our subsequent work, we intend to explore machine learning algorithms and AI-based image analysis tools to improve the accuracy and scalability of our methods. By integrating AI with open-source platforms, we aim to expand the accessibility and applicability of our research for broader use in precision agriculture. The HSB determined from this research will be used to develop a computer vision system for automated quantification of blackberry flowers.

Comment 7: Figure 14, is this FF or PF?

Response 7: Thank you for asking. Figure 14 shows 123 plots for both (75) FF and (47) PF and 1 confidential plots across the study area.

Comment 8: The supplementary table is not downloadable in the review system. Please check it.

Response 8: Thank you for pointing this out.  We have included a descriptive table (Table 2) for the data analyzed and will therefore not be needing the supplementary table.

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Most of the comments were noted and corrected, due to which the authors managed to improve the readability, the general structure of the article and the presentation of the research experiment.

Author Response

Comment 1: Most of the comments were noted and corrected, due to which the authors managed to improve the readability, the general structure of the article and the presentation of the research experiment.

Response 1: Thank you for your feedback and recognition of our efforts to address the comments. We appreciate your insights and perspectives, which significantly contributed to enhancing the readability, structure, and overall presentation of our research. Your input was invaluable in refining our work.

Reviewer 3 Report

Comments and Suggestions for Authors

All questions and comments were answered well.

The presentation quality of figures should be improved, such as the size of axis label.

Author Response

Comment 1: All questions and comments were answered well.

Response 1: Thank you for your feedback. We are pleased to hear that our responses adequately addressed all the questions and comments. Your constructive feedback played a crucial role in improving the quality of our work.

Comment 2: The presentation quality of figures should be improved, such as the size of axis label.

Response 2: Thank you for pointing this out. We have installed and implemented the “ggsave”option from “ggplot2” package in RStudio. This has helped us increase image resolution. We have also increased the axis labels for figures 8 to 13. This has improved the presentation quality of the figures and increased the image resolution to ensure better readability and clarity.