Next Article in Journal
Brain-Inspired Synergistic Adversarial Framework for Style Transfer-Guided Semantic Segmentation in Cross-Domain Remote Sensing Imagery
Previous Article in Journal
Spatial Variation in Coral Diversity and Reef Complexity in the Galápagos: Insights from Underwater Photogrammetry and New Data Extraction Methods
 
 
Article
Peer-Review Record

Early Detection of Pine Wilt Disease by Combining Pigment and Moisture Content Indices Using UAV-Based Hyperspectral Imagery

Remote Sens. 2025, 17(11), 1833; https://doi.org/10.3390/rs17111833
by Rui Hou 1,2, Biyao Zhang 1,*, Guofei Fang 3,4, Sihan Yang 5, Lei Guo 5, Wenjiang Huang 1,2, Jing Yao 1, Quanjun Jiao 1, Hong Sun 3,4 and Jiayu Yan 3,4
Reviewer 1:
Reviewer 2:
Reviewer 3:
Remote Sens. 2025, 17(11), 1833; https://doi.org/10.3390/rs17111833
Submission received: 18 March 2025 / Revised: 12 May 2025 / Accepted: 22 May 2025 / Published: 23 May 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors
  1. INNOVATIVITY AND RESEARCH SIGNIFICANCE
  2. Accurate Problem Orientation

 The thesis proposes a lightweight detection system based on UAV image processing to address the need for efficient detection of pine wilt disease (PWD). Early identification of pine wilt disease (PWD) is crucial for forest protection, while existing models are deficient in balancing lightweight and accuracy. The paper reflects a clear problem orientation by improving the YOLOv5 framework, combining feature fusion and loss function optimization.

  1. Clear technical improvement points

 The EfficientViT building block and CACSNet feature fusion network proposed in the paper reduces the number of model parameters (e.g., the model volume is only 12.8M) and improves the detection accuracy (AP up to 95.2%) through the lightweight multiscale attention mechanism and cross-scale bidirectional fusion strategy. Compared with PWD-YOLO from Nanjing Forestry University (model volume 2.7MB, mAP@0.5è¾¾87.7%), the balance of model lightweight and accuracy is more advantageous.

 

II.EXPERIMENTAL AND DATA SUPPORT

  1. Dataset and Labeling

 The paper does not specify the size of the dataset, labeling standards, and environmental diversity.

   It is suggested to disclose the details of the dataset (e.g., sample size, labeling tools, environmental diversity) and compare the effects of different flight altitudes on the detection accuracy.

  1. Insufficient comparison experiments

 The paper needs to add a comprehensive comparison with mainstream models, and needs to highlight the advantages of this method in a single modality of RGB images.

 

III. Practical application and normative compliance

  1. Technical specification compliance

 The paper does not mention whether it complies with the requirements of UAV monitoring technical standards on flight altitude, resolution, and data processing flow). It is suggested to add:

 - flight parameters (e.g. the relationship between operation altitude and GSD);

 - result output format (e.g. DOM/DEM data generation).

 

  1. Writing and Literature Citation
  2. Insufficient Literature Coverage

The paper does not sufficiently cite the recent key studies, e.g. It is recommended to add relevant literature to clarify the difference between this method and the existing work. 2.

  1. Optimization of presentation

 Some terms need to be unified (e.g., “pine wilt” and “pine wilt disease” are mixed, which should be named according to the standard).

 

  1. Conclusions and Recommendations

The paper is innovative in lightweight model design and accuracy improvement, and provides a feasible solution for pine wilt monitoring by unmanned aerial vehicles, but experimental comparisons, data details, and specification adaptability need to be improved. Key additions:

  1. Quantitative comparison with PWD-YOLO and other models;
  2. Detailed description of dataset construction and labeling methods;
  3. Analysis of technical specification compliance and actual deployment test.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The study addresses an important topic in early detection of pine wilt disease using UAV hyperspectral data, which is highly relevant to forest health monitoring. The approach and results are promising, but several critical issues require clarification and revision:

  1. While the introduction emphasizes developing "new indicators based on PWD pathogenesis," the study instead selects existing VIs without novel development. Additionally, the analysis does not explicitly link results to PWD pathogenesis mechanisms as stated in the objectives. Ithink it is necessary to revise the objectives to align with the actual methodology or expanding the analysis to incorporate pathological insights.
  2. The description of two experimental areas (Figure 1) is inconsistent with the results, where the second area does not appear in Figure 4 and Figure Authors must clarify sample sizes, validation procedures, and whether calibration/validation datasets overlapped. Transparency in data sources is essential for reliability evaluationof the results.
  3. The manuscript lacks details on how leaf-level and UAV-derived canopy spectra were integrated. It is unclear which analyses used leaf vs. image data, how they were matched (e.g., considering leaf position vs. canopy perspective), and whether preprocessing addressed scale differences.
  4. The reliance on pixel-level color thresholds for disease staging is problematic due to: - Natural variations (e.g., senescence, background noise, lighting). - Within-crown spectral heterogeneity (even healthy trees may show variable pixels). The threshold determination process is insufficiently explained. It is essential toquantify uncertainty, incorporating spatial context (e.g., neighboring pixels), or validating with ground-truth data at the pixel scale.
  5. The manuscript does not justify the value of pixel-level detection over tree-level approaches. Authors should discuss trade-offs (e.g., precision vs. noise) and explore whether pixel-scale data provides unique insights for early detection.
  6. Spectral indices were compared across different acquisition dates. Were radiometric corrections applied to account for environmental variability (e.g., weather, sun angle)? This is critical for interpreting temporal trends.
  7. The reference list lacks recent (2023–2024) advancements in PWD hyperspectral monitoring. Some related references should be included to contextualize the work within current knowledge. And also, the References list is disordered.
  8. A one-month sampling interval may miss rapid symptom fluctuations (e.g., transient discoloration followed by recovery). Authors should discuss potential limitations of this temporal resolution and consider incorporating supplementary data (e.g., weekly ground observations) to validate periodicity assumptions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

A well written manuscript  I have very few comments . The study plots could be more completely described.  I have indicated these changes on manuscript. Another suggestion may be to look at your results in relation to the spectrum you developed in figure 3.  Since you have already developed this curve why not try to develop an index specific to Chinese pine.   A technique to develop species specific indices would be useful. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has undergone a comprehensive and meticulous revision according to the review comments, basically addressing the issues and suggestions raised in the initial review. However, although the pixel-level sample extraction can significantly increase the sample quantities, and the authors have explained that necessary measures were taken during the measuring process to minimize errors caused by environmental factors, in the subsequent analysis, especially during threshold segmentation, attention should be paid to the uncertainties that are highly likely to arise.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Back to TopTop