A Novel Framework for Stratified-Coupled BLS Tree Trunk Detection and DBH Estimation in Forests (BSTDF) Using Deep Learning and Optimization Adaptive Algorithm

Round 1

Reviewer 1 Report

This paper introduces the BSTDF framework for detecting forest tree trunks and estimating DBH using BLS. While the proposed method is promising, there are still some concerns and issues that need to be addressed in the next revision.

1. In Figure 8, it is difficult to discern the differences in results between WCF-CACL-RandLA-Net and RandLA-Net. To address this issue, the author should provide some descriptions or annotations to help readers better understand the comparison.

2. The following references need to be cited.

[1] Li, J.T.; Cheng, X.J.; Xiao, Z.H. A branch-trunk-constrained hierarchical clustering method for street trees individual extraction from mobile laser scanning point clouds. Measurement 2022, 189, 110440.

[2] Ning, X.; Ma, Y.; Hou, Y.; Lv, Z.; Jin, H.; Wang, Y. Semantic Segmentation Guided Coarse-to-Fine Detection of Individual Trees from MLS Point Clouds Based on Treetop Points Extraction and Radius Expansion. Remote Sensing. 2022, 14, 4926.

[3] Wang P, Tang Y, Liao Z, Yan Y, Dai L, Liu S, Jiang T. Road-Side Individual Tree Segmentation from Urban MLS Point Clouds Using Metric Learning. Remote Sensing. 2023; 15(8):1992.

3. In the comparison experiments, the author should consider including more comparative methods in addition to RandLA-Net. For example, MS-RRFSegNet,KPConv, Point-attention Net, PointNet, PointNet++, etc.

4. It would be beneficial if the authors could include ablation experiments in their study.

5. Although LSA-RANSAC has shown effectiveness in DBH estimation, it is suggested to compare it with other methods such as Least Squares circle fitting (LS), Cylindrical Fitting, Hough Transform (HT), and Convex Hull Algorithm (CHA).

Author Response

Dear Editor and Reviewers,

We are immensely grateful to the editor and all reviewers for dedicating their time to providing constructive feedback and valuable suggestions. These inputs have significantly enhanced the quality of our manuscript, allowing us to refine it further. We have diligently considered and incorporated each revision suggestion and perspective proposed by the reviewers. Below, we respond to the reviewers' comments, addressing each point individually and indicating corresponding revisions. The entire text has been meticulously reviewed and revised once more. We have uploaded our responses as a Word document. Please see the attachment.

Best Regards,

Huacong zhang

Author Response File: Author Response.docx

Reviewer 2 Report

The study demonstrated the effectiveness of BSTDF in forest DBH estimation, offering a more efficient solution for forest resource monitoring and quantification, and possessing immense potential to replace field forest measurements.However, the paper still has the following deficiencies.

1.Line 142， Table 1 should describe the data characteristics of validation data and modeling data separately.

2.Line 335， Don’t forget to cite the software!  For example in section of 2.3 “Matlab R2021a”.

3.In the section of 4,the discussion part is not sufficiently written. It should be compared more deeply with the results of other scholars.

The author's English writing is relatively standardized, however,individual statements need to be modified.

Author Response

Dear Editor and Reviewers,

We are immensely grateful to the editor and all reviewers for dedicating their time to providing constructive feedback and valuable suggestions. These inputs have significantly enhanced the quality of our manuscript, allowing us to refine it further. We have diligently considered and incorporated each revision suggestion and perspective proposed by the reviewers. Below, we respond to the reviewers' comments, addressing each point individually and indicating corresponding revisions. The entire text has been meticulously reviewed and revised once more. We have uploaded our responses as a Word document. Please see the attachment.

Best Regards,

Huacong zhang

Author Response File: Author Response.docx

Reviewer 3 Report

The paper develops a novel framework for stem detection and DBH estimation. The framework creates a stem detection deep learning dataset using a stratified coupling approach, which introduces a weighted cross-entropy focal-loss function module and a cosine annealing cyclic learning strategy. Meanwhile, the DBH is estimated using an LSA-RANSAC cylindrical fitting method. Overall, this study has high innovation for stem detection and DBH estimation. It is are recommended to be published after minor revisions. I only have some minor concerns:

(1)   Shorten the title, with as few abbreviations as possible.

(2)   In Line 26, please provide the full name of LSA-RANSAC.

(3)   In Lines 99-102, the contribution of this study needs to be described in detail.

(4)   The information of BLS can be introduced more intuitively, if its images are provided.

(5)   The resolution of Figures 1d and 11 needs to be improved.

(6)   In Lines 352-357, all abbreviations should be italicized.

Author Response

Dear Editor and Reviewers,

We are immensely grateful to the editor and all reviewers for dedicating their time to providing constructive feedback and valuable suggestions. These inputs have significantly enhanced the quality of our manuscript, allowing us to refine it further. We have diligently considered and incorporated each revision suggestion and perspective proposed by the reviewers. Below, we respond to the reviewers' comments, addressing each point individually and indicating corresponding revisions. The entire text has been meticulously reviewed and revised once more. We have uploaded our responses as a Word document. Please see the attachment.

Best Regards,

Huacong zhang

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The author has provided detailed revisions one by one according to the review comments, and I have no further suggestions for revisions.