pith. sign in

arxiv: 2606.08206 · v2 · pith:XJHAY6M3new · submitted 2026-06-06 · 💻 cs.CV · cs.LG

SegmentAnyTreeV2: Scaling Transformer-Based Tree Instance Segmentation Across Sensors, Platforms, and Forests

Maciej Wielgosz , Stefano Puliti , Rasmus Astrup This is my paper

Pith reviewed 2026-06-27 19:37 UTC · model grok-4.3

classification 💻 cs.CV cs.LG
keywords tree instance segmentationpoint cloudLiDARtransformerforestsemantic segmentationFOR-instance datasetcross-attention decoder
0
0 comments X

The pith

SegmentAnyTreeV2 segments individual trees from any LiDAR sensor or platform using a Point Transformer backbone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces SegmentAnyTreeV2 as a framework that performs both semantic and instance segmentation on forest point clouds without depending on specific sensors or flying platforms. It pairs a serialization-based Point Transformer v3 with a semantic head that limits instance decoding to tree voxels and a cross-attention mask decoder that uses instance-aware query initialization and one-to-many supervision. The authors also release FOR-instance v3, an expanded collection of 427 scenes containing over 26,000 annotated trees spanning multiple biomes and LiDAR types. On the FOR-instanceV2 test split the model records 90.5 percent precision, 80.2 percent recall, 85.0 percent F1, 90.7 percent coverage and 87.6 percent semantic mIoU while maintaining performance in zero-shot tests on unseen sites.

Core claim

SegmentAnyTreeV2 is a sensor- and platform-agnostic framework for semantic and instance segmentation of forest point clouds that combines a serialization-based Point Transformer v3 backbone with a lightweight semantic head and a tree-focused cross-attention mask decoder; semantic predictions restrict instance decoding to tree-class voxels, while instance-aware query initialization, one-to-many seed supervision, and asymmetric mask scoring improve separation in dense stands, yielding 90.5 percent precision, 80.2 percent recall, 85.0 percent F1, 90.7 percent coverage, and 87.6 percent semantic mIoU on the FOR-instanceV2 test split with strong zero-shot cross-domain generalization.

What carries the argument

SegmentAnyTreeV2 framework, which restricts instance decoding to tree voxels via a semantic head and applies cross-attention mask decoding with instance-aware queries to separate trees in point clouds.

If this is right

  • Automated tree detection and delineation become feasible at scale across mixed forest types and LiDAR sources.
  • Inventory and carbon accounting pipelines can operate on point clouds collected from drones, helicopters, or ground platforms without retraining.
  • Instance masks enable downstream tasks such as individual-tree biomass estimation from the same input data.
  • Zero-shot transfer reduces the need for site-specific annotation when moving to new regions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach could extend to multi-temporal LiDAR sequences for tracking individual tree growth or mortality.
  • Integration with optical imagery might allow hybrid models that combine structural and spectral cues for species identification.
  • The serialization and cross-attention design offers a template for instance segmentation of other natural objects such as rocks or crops in 3-D data.

Load-bearing premise

The FOR-instance v3 dataset captures a representative range of real forest conditions, sensor types, and platforms so that measured performance will hold on new data.

What would settle it

Running the released model on a fresh LiDAR collection from a biome absent from FOR-instance v3 and recording whether F1 falls below 70 percent would test the generalization claim.

Figures

Figures reproduced from arXiv: 2606.08206 by Maciej Wielgosz, Rasmus Astrup, Stefano Puliti.

Figure 1
Figure 1. Figure 1: Overview of the proposed architecture. A cylindrical crop is voxelized and processed by the PTv3 backbone. The resulting per-voxel features feed a [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visualization of the structural complexity and forest type range within FOR-instance V3. [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Structural diversity of FOR-instance v3 (26 496 instances, 427 scenes) compared to FOR-instance v2 (orange, semi-transparent; 10 597 instances, 93 [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Evolution of state-of-the-art performance across the FOR-instance dataset versions, measured by macro-F1 score. The figure also indicates the corresponding [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Examples of undersegmentation of nearby trees with deeply inter [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Examples of performance in Robson Creek with details on correctly segmented points (green), false positives (red), and false negatives (blue) [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
read the original abstract

We present SegmentAnyTreeV2, a sensor- and platform-agnostic framework for semantic and instance segmentation of forest point clouds. The model combines a serialization-based Point Transformer v3 backbone with a lightweight semantic head and a tree-focused cross-attention mask decoder. Semantic predictions restrict instance decoding to tree-class voxels, while instance-aware query initialization, one-to-many seed supervision, and asymmetric mask scoring improve separation in dense and structurally complex stands. We further introduce FOR-instance v3, an expanded benchmark comprising 427 scenes and 26,496 annotated trees across diverse biomes, forest structures, and LiDAR platforms. On the FOR-instanceV2 test split, SegmentAnyTreeV2 achieves 90.5% precision, 80.2% recall, 85.0% F1, 90.7% coverage, and 87.6% semantic mIoU, outperforming previous learning-based methods in both instance detection and mask completeness. Zero-shot evaluation on independent sites further demonstrates strong cross-domain generalization.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript presents SegmentAnyTreeV2, a sensor- and platform-agnostic framework for semantic and instance segmentation of forest point clouds. It uses a serialization-based Point Transformer v3 backbone with a lightweight semantic head and a tree-focused cross-attention mask decoder, incorporating semantic predictions to restrict instance decoding, instance-aware query initialization, one-to-many seed supervision, and asymmetric mask scoring. The authors introduce FOR-instance v3, an expanded benchmark with 427 scenes and 26,496 annotated trees across diverse biomes, structures, and LiDAR platforms. On the FOR-instanceV2 test split, the model achieves 90.5% precision, 80.2% recall, 85.0% F1, 90.7% coverage, and 87.6% semantic mIoU, outperforming prior learning-based methods, with additional zero-shot evaluation on independent sites demonstrating cross-domain generalization.

Significance. If the reported metrics are supported by a properly disjoint train/test split and full experimental details, the work would advance the field by scaling transformer-based tree segmentation to diverse real-world conditions and providing a substantially larger benchmark dataset. The combination of semantic restriction and specialized instance decoding addresses challenges in dense stands, and the zero-shot results would support claims of sensor/platform agnosticism.

major comments (2)
  1. [Abstract] Abstract and results paragraph: the headline metrics (90.5% precision, 80.2% recall, 85.0% F1 on FOR-instanceV2 test split) are presented as evidence of superiority and generalization, yet the manuscript does not state whether FOR-instanceV2 test scenes were excluded from the FOR-instance v3 training set. Overlap would mean the numbers reflect memorization rather than the claimed sensor- and platform-agnostic performance.
  2. [Abstract] Abstract: specific performance numbers are reported without any information on experimental setup, including training data composition, baselines, statistical significance, error bars, or potential data issues, preventing verification that the data supports the central claims.
minor comments (1)
  1. The terms 'one-to-many seed supervision' and 'asymmetric mask scoring' are mentioned in the abstract but require explicit definitions and algorithmic details in the methods section for reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback highlighting the need for clearer statements on data splits and experimental details. We address both major comments below and will incorporate clarifications in a revised manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract and results paragraph: the headline metrics (90.5% precision, 80.2% recall, 85.0% F1 on FOR-instanceV2 test split) are presented as evidence of superiority and generalization, yet the manuscript does not state whether FOR-instanceV2 test scenes were excluded from the FOR-instance v3 training set. Overlap would mean the numbers reflect memorization rather than the claimed sensor- and platform-agnostic performance.

    Authors: The FOR-instanceV2 test scenes are fully excluded from the FOR-instance v3 training set. FOR-instance v3 augments the prior v2 data with new scenes across additional biomes and platforms while preserving the original v2 test split as a completely held-out evaluation set. This design directly supports the sensor- and platform-agnostic claims. We will add an explicit sentence in the abstract and expand the dataset description section to state the disjoint split. revision: yes

  2. Referee: [Abstract] Abstract: specific performance numbers are reported without any information on experimental setup, including training data composition, baselines, statistical significance, error bars, or potential data issues, preventing verification that the data supports the central claims.

    Authors: The abstract is intentionally concise, but the full manuscript details the training composition of FOR-instance v3, the full set of baselines, and reports error bars plus per-scene metrics in the experimental section and tables. We will revise the abstract to briefly note the disjoint test split, the comparison against prior learning-based methods, and direct readers to the methods and results sections for complete experimental protocols, including any data preprocessing steps. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical results on held-out splits and zero-shot sites

full rationale

The paper introduces FOR-instance v3 as an expanded benchmark and reports metrics on the FOR-instanceV2 test split plus zero-shot evaluation on independent sites. These are standard held-out empirical evaluations with no self-definitional equations, fitted parameters renamed as predictions, or load-bearing self-citations that reduce the central claims to inputs by construction. The architecture description (Point Transformer v3 backbone, semantic head, mask decoder) is presented as design choices, not derived from the reported numbers. No patterns from the enumerated circularity kinds are present.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Only abstract available so ledger is minimal; full paper would likely list more training details and assumptions about data distribution.

free parameters (1)
  • Various model hyperparameters
    Deep learning models like this typically have many hyperparameters tuned on data, but specifics not in abstract.
axioms (1)
  • domain assumption Point Transformer v3 is an appropriate backbone for processing forest point clouds.
    The model builds directly on this architecture as stated in the abstract.

pith-pipeline@v0.9.1-grok · 5718 in / 1456 out tokens · 43517 ms · 2026-06-27T19:37:54.949283+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. SelectAnyTree: A Promptable Instance Segmentation Model for 3D Forest LiDAR Point Clouds

    cs.CV 2026-06 conditional novelty 7.0

    SelectAnyTree is a promptable instance segmentation model for 3D forest LiDAR point clouds that achieves 78.2 IoU from a single click via a click-to-query prompt encoder, CHM-guided first prompt, and state-space query...

Reference graph

Works this paper leans on

124 extracted references · 13 canonical work pages · cited by 1 Pith paper

  1. [1]

    2019 , pages =

    Choy, Christopher and Gwak, JunYoung and Savarese, Silvio , booktitle =. 2019 , pages =

    2019

  2. [2]

    Zamir and Silvio Savarese , journal =

    Iro Armeni and Sasha Sax and Amir R. Zamir and Silvio Savarese , journal =. Joint

  3. [3]

    Kuhn, Harold W , journal =. The. 1955 , number =

    1955

  4. [4]

    2023 , pages =

    Schult, Jonas and Engelmann, Francis and Hermans, Alexander and Litany, Or and Tang, Siyu and Leibe, Bastian , booktitle =. 2023 , pages =

    2023

  5. [5]

    Andreas Tockner and Christoph Gollob and Tim Ritter and Arne Nothdurft , doi =

  6. [6]

    Remote Sensing , VOLUME =

    Peng, Xiangji and Yi, Jizheng and Liu, Rong and Shen, Xiangyu and Li, Xiaoyao , TITLE =. Remote Sensing , VOLUME =. 2026 , NUMBER =

    2026

  7. [7]

    NIPS , title =

    Yuchen Bai and Jean. NIPS , title =. 2023 , pages =

    2023

  8. [8]

    Scientific Data , year =

    Stereńczak, Krzysztof and Kulicki, Maksymilian and Kraszewski, Bartłomiej and Gasica, Torana Arya and Mitelsztedt, Krzysztof and Mielcarek, Miłosz and Tangwa, Elvis and Erfanifard, Yousef and Brach, Michał and others , title =. Scientific Data , year =. doi:10.1038/s41597-026-07269-1 , url =

    work page doi:10.1038/s41597-026-07269-1

  9. [9]

    Query Refinement Transformer for

    Lu, Jiahao and Deng, Jiacheng and Wang, Chuxin and He, Jianfeng and Zhang, Tianzhu , booktitle =. Query Refinement Transformer for. 2023 , pages =

    2023

  10. [10]

    2024 , pages =

    Kolodiazhnyi, Maxim and Vorontsova, Anna and Konushin, Anton and Rukhovich, Danila , booktitle =. 2024 , pages =

    2024

  11. [11]

    2023 , pages =

    He, Junjie and Li, Pengyu and Geng, Yifeng and Xie, Xuansong , booktitle =. 2023 , pages =

    2023

  12. [12]

    Liu, Jiaheng and He, Tong and Yang, Honghui and Su, Rui and Tian, Jiayi and Wu, Junran and Guo, Hongcheng and Xu, Ke and Ouyang, Wanli , journal =

  13. [13]

    Semantic Instance Segmentation for Autonomous Driving , year =

    De Brabandere, Bert and Neven, Davy and Van Gool, Luc , booktitle =. Semantic Instance Segmentation for Autonomous Driving , year =. doi:10.1109/CVPRW.2017.66 , keywords =

    work page doi:10.1109/cvprw.2017.66 2017

  14. [14]

    Secretariat of the Convention on Biological Diversity , title =

  15. [15]

    Intergovernamental Panel on Climate Change (IPCC) , title =

  16. [16]

    Towards accurate instance segmentation in large-scale

    Binbin Xiang and Torben Peters and Theodora Kontogianni and Frawa Vetterli and Stefano Puliti and Rasmus Astrup and Konrad Schindler , journal =. Towards accurate instance segmentation in large-scale. 2023 , pages =

    2023

  17. [17]

    Superpoint transformer for

    Sun, Jiahao and Qing, Chunmei and Tan, Junpeng and Xu, Xiangmin , booktitle =. Superpoint transformer for

  18. [18]

    2023 , pages =

    Zong, Zhuofan and Song, Guanglu and Liu, Yu , booktitle =. 2023 , pages =

    2023

  19. [19]

    2018 , pages =

    Wang, Weiyue and Yu, Ronald and Huang, Qiangui and Neumann, Ulrich , booktitle =. 2018 , pages =

    2018

  20. [20]

    Improved Block Merging for

    Denis, Leon and Royen, Remco and Munteanu, Adrian , booktitle =. Improved Block Merging for. 2023 , pages =

    2023

  21. [21]

    Stefano Puliti and Grant Pearse and Peter Surový and Luke Wallace and Markus Hollaus and Maciej Wielgosz and Rasmus Astrup , journal =

  22. [22]

    2020 , pages =

    Jiang, Li and Zhao, Hengshuang and Shi, Shaoshuai and Liu, Shu and Fu, Chi-Wing and Jia, Jiaya , booktitle =. 2020 , pages =

    2020

  23. [23]

    2024 , pages =

    Maciej Wielgosz and Stefano Puliti and Binbin Xiang and Konrad Schindler and Rasmus Astrup , journal =. 2024 , pages =

    2024

  24. [24]

    Mark Danson and Miro Demol and Mathias Disney and Rachel Gaulton and Sruthi M

    Kim Calders and Jennifer Adams and John Armston and Harm Bartholomeus and Sebastien Bauwens and Lisa Patrick Bentley and Jerome Chave and F. Mark Danson and Miro Demol and Mathias Disney and Rachel Gaulton and Sruthi M. Remote Sensing of Environment , title =. 2020 , pages =

    2020

  25. [25]

    ITS-Net: A platform and sensor agnostic 3D deep learning model for individual tree segmentation using aerial LiDAR data , journal =

    Bowen Li and Yong Pang and Daniel Kükenbrink and Luo Wang and Dan Kong and Mauro Marty , keywords =. ITS-Net: A platform and sensor agnostic 3D deep learning model for individual tree segmentation using aerial LiDAR data , journal =. 2026 , issn =. doi:10.1016/j.isprsjprs.2025.11.019 , url =

    work page doi:10.1016/j.isprsjprs.2025.11.019 2026

  26. [26]

    Geo-spatial Information Science , volume =

    Xinlian Liang and Hanwen Qi and Xuejie Deng and Jianchang Chen and Shangshu Cai and Qingjun Zhang and Yunsheng Wang and Antero Kukko and Juha Hyyppä , title =. Geo-spatial Information Science , volume =. 2025 , publisher =

    2025

  27. [27]

    2025 , month = jan, version =

    Puletti, Nicola and Guasti, Matteo and Innocenti, Simone and Botticelli, Davide , title =. 2025 , month = jan, version =. doi:10.5281/zenodo.14687890 , url =

    work page doi:10.5281/zenodo.14687890 2025

  28. [28]

    2024 , publisher =

    Anonymous , title =. 2024 , publisher =. doi:10.5281/zenodo.13268500 , url =

    work page doi:10.5281/zenodo.13268500 2024

  29. [29]

    and Takhtkeshha, N

    Bocaux, L. and Takhtkeshha, N. and Ma, Z. and Remondino, F. , TITLE =. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences , VOLUME =. 2025 , PAGES =

    2025

  30. [30]

    2025 , publisher =

    Xiang, Binbin and Wielgosz, Maciej and Puliti, Stefano and Král, Kamil and Krucek, Martin and Missarov, Azim and Astrup, Rasmus , title =. 2025 , publisher =. doi:10.5281/zenodo.16742708 , url =

    work page doi:10.5281/zenodo.16742708 2025

  31. [31]

    2026 , issn =

    Cross-platform forest understanding: A multi-platform synergistic training framework for generalized forest point cloud segmentation , journal =. 2026 , issn =. doi:10.1016/j.rse.2026.115467 , url =

    work page doi:10.1016/j.rse.2026.115467 2026

  32. [32]

    2023 , eprint=

    Towards accurate instance segmentation in large-scale LiDAR point clouds , author=. 2023 , eprint=

    2023

  33. [33]

    and Jiang, Helen and Brilakis, Ioannis and Fischer, Martin and Savarese, Silvio , booktitle =

    Armeni, Iro and Sener, Ozan and Zamir, Amir R. and Jiang, Helen and Brilakis, Ioannis and Fischer, Martin and Savarese, Silvio , booktitle =. 2016 , pages =

    2016

  34. [34]

    and Savva, Manolis and Halber, Maciej and Funkhouser, Thomas and Nie

    Dai, Angela and Chang, Angel X. and Savva, Manolis and Halber, Maciej and Funkhouser, Thomas and Nie. CVPR , title =. 2017 , pages =

    2017

  35. [35]

    2019 , pages =

    Behley, Jens and Garbade, Martin and Milioto, Andres and Quenzel, Jan and Behnke, Sven and Stachniss, Cyrill and Gall, J\"urgen , booktitle =. 2019 , pages =

    2019

  36. [36]

    Lang and Sourabh Vora and Venice Erin Liong and Qiang Xu and Anush Krishnan and Yu Pan and Giancarlo Baldan and Oscar Beijbom , booktitle =

    Holger Caesar and Varun Bankiti and Alex H. Lang and Sourabh Vora and Venice Erin Liong and Qiang Xu and Anush Krishnan and Yu Pan and Giancarlo Baldan and Oscar Beijbom , booktitle =. nuScenes:. 2020 , pages =

    2020

  37. [37]

    Hierarchical Aggregation for

    Chen, Shaoyu and Fang, Jiemin and Zhang, Qian and Liu, Wenyu and Wang, Xinggang , booktitle =. Hierarchical Aggregation for. 2021 , pages =

    2021

  38. [38]

    and Nguyen, Thanh and Yoo, Chang D

    Vu, Thang and Kim, Kookhoi and Luu, Tung M. and Nguyen, Thanh and Yoo, Chang D. , booktitle =. 2022 , pages =

    2022

  39. [39]

    2022 , pages =

    Zhong, Min and Chen, Xinghao and Chen, Xiaokang and Zeng, Gang and Wang, Yunhe , booktitle =. 2022 , pages =

    2022

  40. [40]

    Associatively Segmenting Instances and Semantics in Point Clouds , year =

    Xinlong Wang and Shu Liu and Xiaoyong Shen and Chunhua Shen and Jiaya Jia , booktitle =. Associatively Segmenting Instances and Semantics in Point Clouds , year =

  41. [41]

    CVPR , title =

    Engelmann, Francis and Bokeloh, Martin and Fathi, Alireza and Leibe, Bastian and Nie. CVPR , title =. 2020 , pages =

    2020

  42. [42]

    Learning Object Bounding Boxes for

    Bo Yang and Jianan Wang and Ronald Clark and Qingyong Hu and Sen Wang and Andrew Markham and Niki Trigoni , booktitle =. Learning Object Bounding Boxes for. 2019 , pages =

    2019

  43. [43]

    Top-Down Beats Bottom-Up in

    Kolodiazhnyi, Maksim and Vorontsova, Anna and Konushin, Anton and Rukhovich, Danila , booktitle =. Top-Down Beats Bottom-Up in. 2024 , pages =

    2024

  44. [44]

    and Prisacariu, Victor , booktitle =

    Zhang, Feihu and Guan, Chenye and Fang, Jin and Bai, Song and Yang, Ruigang and Torr, Philip H.S. and Prisacariu, Victor , booktitle =. Instance Segmentation of. 2020 , pages =

    2020

  45. [45]

    and Su, Hao and Mo, Kaichun and Guibas, Leonidas J

    Qi, Charles R. and Su, Hao and Mo, Kaichun and Guibas, Leonidas J. , booktitle =. 2017 , pages =

    2017

  46. [46]

    and Yi, Li and Su, Hao and Guibas, Leonidas J

    Qi, Charles R. and Yi, Li and Su, Hao and Guibas, Leonidas J. , booktitle =. 2017 , pages =

    2017

  47. [47]

    and Deschaud, Jean-Emmanuel and Marcotegui, Beatriz and Goulette, François and Guibas, Leonidas , booktitle =

    Thomas, Hugues and Qi, Charles R. and Deschaud, Jean-Emmanuel and Marcotegui, Beatriz and Goulette, François and Guibas, Leonidas , booktitle =. 2019 , pages =

    2019

  48. [48]

    Chen Liu and Yasutaka Furukawa , journal =

  49. [49]

    2019 , pages =

    Pham, Quang-Hieu and Nguyen, Thanh and Hua, Binh-Son and Roig, Gemma and Yeung, Sai-Kit , booktitle =. 2019 , pages =

    2019

  50. [50]

    arXiv , timestamp =

    Cathrin Elich and Francis Engelmann and Theodora Kontogianni and Bastian Leibe , booktitle =. arXiv , timestamp =. 2019 , pages =

    2019

  51. [51]

    Occuseg: Occupancy-aware

    Han, Lei and Zheng, Tian and Xu, Lan and Fang, Lu , booktitle =. Occuseg: Occupancy-aware. 2020 , pages =

    2020

  52. [52]

    Yi and Wang Zhao and He Wang and Minhyuk Sung and L

    L. Yi and Wang Zhao and He Wang and Minhyuk Sung and L. Guibas , booktitle =. 2019 , pages =

    2019

  53. [53]

    Lei Yao and Yi Wang and Moyun Liu and Lap-Pui Chau , journal =

  54. [54]

    Spherical Mask: Coarse-to-Fine

    Sangyun Shin and Kaichen Zhou and Madhu Vankadari and Andrew Markham and Niki Trigoni , booktitle =. Spherical Mask: Coarse-to-Fine. 2024 , pages =

    2024

  55. [55]

    Learnable Earth Parser: Discovering

    Romain Loiseau and Elliot Vincent and Mathieu Aubry and Loic Landrieu , booktitle =. Learnable Earth Parser: Discovering. 2024 , pages =

    2024

  56. [56]

    A Novel Approach for the Detection of Standing Tree Stems from Plot-Level Terrestrial Laser Scanning Data , year =

    Zhang, Wuming and Wan, Peng and Wang, Tiejun and Cai, Shangshu and Chen, Yiming and Jin, Xiuliang and Yan, Guangjian , journal =. A Novel Approach for the Detection of Standing Tree Stems from Plot-Level Terrestrial Laser Scanning Data , year =

  57. [57]

    and Kelle, O

    Hyyppa, J. and Kelle, O. and Lehikoinen, M. and Inkinen, M. , journal =. A segmentation-based method to retrieve stem volume estimates from. 2001 , number =

    2001

  58. [58]

    Individual tree detection and estimation of stem attributes with mobile laser scanning along boreal forest roads , year =

    Raul de Paula Pires and Kenneth Olofsson and Henrik Jan Persson and Eva Lindberg and Johan Holmgren , journal =. Individual tree detection and estimation of stem attributes with mobile laser scanning along boreal forest roads , year =

  59. [59]

    Methods in Ecology and Evolution , volume =

    Burt, Andrew and Disney, Mathias and Calders, Kim , title =. Methods in Ecology and Evolution , volume =. doi:https://doi.org/10.1111/2041-210X.13121 , url =. https://besjournals.onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.13121 , abstract =

    work page doi:10.1111/2041-210x.13121 2041

  60. [60]

    Kershaw Jr

    Elias Ayrey and Shawn Fraver and John A. Kershaw Jr. and Laura S. Kenefic and Daniel Hayes and Aaron R. Weiskittel and Brian E. Roth , journal =. Layer Stacking: A Novel Algorithm for Individual Forest Tree Segmentation from. 2017 , number =

    2017

  61. [61]

    , title =

    Dalponte, Michele and Coomes, David A. , title =. Methods in Ecology and Evolution , volume =. 2016 , doi =

    2016

  62. [62]

    arXiv preprint arXiv:2503.04420 , year=

    PointsToWood: A deep learning framework for complete canopy leaf-wood segmentation of TLS data across diverse European forests , author=. arXiv preprint arXiv:2503.04420 , year=

    work page arXiv

  63. [63]

    Calders, Kim and Verbeeck, Hans and Burt, Andrew and Origo, Niall and Nightingale, Joanne and Malhi, Yadvinder and Wilkes, Phil and Raumonen, Pasi and Bunce, Robert G. H. and Disney, Mathias , title =. Ecological Solutions and Evidence , volume =. 2022 , doi =

    2022

  64. [64]

    Estimating plot-level tree heights with lidar: local filtering with a canopy-height based variable window size , year =

    Sorin C Popescu and Randolph H Wynne and Ross F Nelson , journal =. Estimating plot-level tree heights with lidar: local filtering with a canopy-height based variable window size , year =

  65. [65]

    Remote Sensing of Environment , title =

    Ant\'onio Ferraz and Fr\'ed\'eric Bretar and St\'ephane Jacquemoud and Gil Gon. Remote Sensing of Environment , title =. 2012 , pages =

    2012

  66. [66]

    A tree detection method based on trunk point cloud section in dense plantation forest using drone

    Yupan Zhang and Yiliu Tan and Yuichi Onda and Asahi Hashimoto and Takashi Gomi and Chenwei Chiu and Shodai Inokoshi , journal =. A tree detection method based on trunk point cloud section in dense plantation forest using drone. 2023 , pages =

    2023

  67. [67]

    Individual tree identification using a new cluster-based approach with discrete-return airborne

    Haijian Liu and Pinliang Dong and Changshan Wu and Pin Wang and Meihong Fang , journal =. Individual tree identification using a new cluster-based approach with discrete-return airborne. 2021 , pages =

    2021

  68. [68]

    Remote Sensing , VOLUME =

    Xi, Zhouxin and Hopkinson, Chris , TITLE =. Remote Sensing , VOLUME =. 2022 , NUMBER =

    2022

  69. [69]

    and Scholes, Robert J

    Sankaran, Mahesh and Hanan, Niall P. and Scholes, Robert J. and Ratnam, Jayashree and Augustine, David J. and Cade, Brian S. and Gignoux, Jacques and Higgins, Steven I. and Le Roux, Xavier and Ludwig, Fulco and Ard. Determinants of Woody Cover in African Savannas , journal =. 2005 , volume =

    2005

  70. [70]

    RayExtract: A fast, scalable method for tree volume reconstruction from terrestrial laser scanning , journal =

    Timothy Devereux and Thomas Lowe and Joshua Rivory and Rafael. RayExtract: A fast, scalable method for tree volume reconstruction from terrestrial laser scanning , journal =. 2026 , issn =. doi:10.1016/j.rse.2025.115162 , url =

    work page doi:10.1016/j.rse.2025.115162 2026

  71. [71]

    and Disney, Mathias and Wilkes, Phil and Burt, Andrew and Calders, Kim and Woodgate, William , title =

    Vicari, Matheus B. and Disney, Mathias and Wilkes, Phil and Burt, Andrew and Calders, Kim and Woodgate, William , title =. Methods in Ecology and Evolution , volume =. 2019 , doi =

    2019

  72. [72]

    Victor F. Str. ISPRS Journal of Photogrammetry and Remote Sensing , title =. 2015 , pages =

    2015

  73. [73]

    Eichhorn and Jiangchuan Fan and Fangzhou Liu and Bangqian Chen and Feng An and Lin Cao , journal =

    Ting Yun and Kang Jiang and Guangchao Li and Markus P. Eichhorn and Jiangchuan Fan and Fangzhou Liu and Bangqian Chen and Feng An and Lin Cao , journal =. Individual tree crown segmentation from airborne. 2021 , pages =

    2021

  74. [74]

    Towards low vegetation identification: A new method for tree crown segmentation from

    Langning Huo and Eva Lindberg and Johan Holmgren , journal =. Towards low vegetation identification: A new method for tree crown segmentation from. 2022 , pages =

    2022

  75. [75]

    2020 , number =

    Wang, Di and Momo Takoudjou, Stéphane and Casella, Eric , journal =. 2020 , number =

    2020

  76. [76]

    Separation of Wood and Foliage for Trees From Ground Point Clouds Using a Novel Least-Cost Path Model , year =

    Xu, Sheng and Zhou, Kai and Sun, Yuan and Yun, Ting , journal =. Separation of Wood and Foliage for Trees From Ground Point Clouds Using a Novel Least-Cost Path Model , year =

  77. [77]

    A novel and efficient method for wood–leaf separation from terrestrial laser scanning point clouds at the forest plot level , year =

    Peng Wan and Jie Shao and Shuangna Jin and Tiejun Wang and Shengmei Yang and Guangjian Yan and Wuming Zhang , journal =. A novel and efficient method for wood–leaf separation from terrestrial laser scanning point clouds at the forest plot level , year =

  78. [78]

    Filtering Stems and Branches from Terrestrial Laser Scanning Point Clouds Using Deep

    Xi, Zhouxin and Hopkinson, Chris and Chasmer, Laura , journal =. Filtering Stems and Branches from Terrestrial Laser Scanning Point Clouds Using Deep. 2018 , number =

    2018

  79. [79]

    Individual Tree Crown Segmentation Directly from

    Chen, Xinxin and Jiang, Kang and Zhu, Yushi and Wang, Xiangjun and Yun, Ting , journal =. Individual Tree Crown Segmentation Directly from. 2021 , number =

    2021

  80. [80]

    Sensor Agnostic Semantic Segmentation of Structurally Diverse and Complex Forest Point Clouds Using Deep Learning , year =

    Krisanski, Sean and Taskhiri, Mohammad Sadegh and Gonzalez Aracil, Susana and Herries, David and Turner, Paul , journal =. Sensor Agnostic Semantic Segmentation of Structurally Diverse and Complex Forest Point Clouds Using Deep Learning , year =

Showing first 80 references.