canopy height estimation

This study addresses the limitation of existing remote sensing–based tree height estimation methods, which predominantly rely on point predictions and lack quantification of uncertainty, thereby hindering their utility in risk-sensitive ecological decision-making. To overcome this, the work proposes the first integration of quantile regression into a lightweight deep learning framework for tree height estimation, enabling the generation of statistically calibrated uncertainty intervals directly from satellite imagery. The approach not only achieves efficient uncertainty modeling but also reveals meaningful relationships between model confidence and key remote sensing challenges—such as terrain complexity and vegetation heterogeneity—automatically yielding lower confidence in more complex environments. Experimental results demonstrate that the proposed method significantly enhances the reliability and applicability of remotely sensed tree height products.

This study addresses the lack of high-resolution, fine-scale quantification of urban tree biomass, which hinders the characterization of individual-tree heterogeneity. The authors propose a self-supervised dual-stream cross-attention network that fuses airborne LiDAR with near-infrared RGB imagery to generate semantic labels, enabling annotation-free crown delineation through multiscale watershed segmentation. Aboveground biomass is then estimated using species-specific allometric equations. The work introduces the first publicly available bitemporal non-forest tree biomass database and incorporates deep ensemble uncertainty maps to guide model refinement. On an independent test set, biomass predictions achieve R² values of 0.570–0.609. Applied to an 810 km² area in Ontario from 2018 to 2023, the approach reveals a net carbon stock increase of 39 Gg C, with localized densities reaching up to 140 Mg/ha.

This study addresses the challenge of outdated urban canopy data by proposing an optical GeoAI framework integrating DeepForest and SAM to assess tree canopy coverage and its thermal effects in Davis, California. The method generates transparent attention surfaces that ensure both data traceability and spatial diagnostics. Experimental mapping of 2.43 km² achieved a 97.4% centroid agreement with LiDAR references, confirming a significant negative correlation between canopy cover and land surface temperature while revealing community-level spatial structures. This research provides urban planning with a high-precision, reproducible screening layer that effectively complements existing remote sensing products, offering a robust solution for timely urban environmental monitoring and heat mitigation strategies.

A lack of high-quality, multi-temporal, multispectral, and LiDAR-derived canopy height model (CHM) co-registered open datasets hinders sub-meter tree-height prediction. Method: We introduce PrediTree—the first open, sub-meter (0.5 m), multi-temporal, multispectral image–CHM paired dataset covering diverse forest ecosystems across France, comprising 3.14 million samples—and propose a U-Net-based encoder–decoder architecture with explicit temporal awareness, jointly modeling multi-temporal multispectral imagery and inter-image time-difference cues for end-to-end CHM regression. Contribution/Results: Trained on PrediTree, our model achieves a masked mean squared error of 11.78%, outperforming ResNet-50 by ~12% and a single-RGB baseline by ~30%. This work bridges critical gaps in both high-resolution forest structural dynamics modeling—providing the first large-scale, temporally aligned benchmark—and methodological capability for fine-grained, time-aware CHM prediction.

To address the challenge of modeling dynamic plant growth in controlled environments, this paper systematically reviews and innovatively integrates deterministic, probabilistic, and generative modeling paradigms into a domain-informed, data-driven framework for forecasting growth under dynamic uncertainty. Methodologically, it unifies spatiotemporal trait evolution with environmental interaction mechanisms, combining regression models, deep neural networks, functional–structural plant models (FSPMs), and conditional generative models (GANs/VAEs), tailored to 2D/3D structured phenotypic data. For the first time, it rigorously characterizes the performance boundaries and applicability domains of each approach. The resulting predictive system achieves interpretability, robustness, and environment-responsive adaptability—overcoming limitations of static experimental paradigms. This work establishes a methodological foundation for high-throughput phenotyping and intelligent, real-world agricultural control.

This study addresses the challenges of instance segmentation of broadleaf tree crowns, which suffer from high morphological variability and ambiguous crown boundaries, leading to limited accuracy and poor generalization. Leveraging a dataset of 18,507 high-quality manually annotated crown polygons, we develop a deep learning framework centered on Mask2Former with various backbone networks, trained and evaluated exclusively on UAV-derived RGB orthoimagery. Our approach achieves high-detail, generalizable individual tree crown segmentation across diverse ecological settings—including temperate forests in Japan and tropical rainforests in Borneo—demonstrating for the first time the critical role of large-scale, fine-grained annotations in enhancing model generalization. The trained model has been integrated into DF Scanner Pro software to support operational forest monitoring applications.

Existing spaceborne LiDAR observations of forest canopy height are limited by spatial sparsity and uncertainty, hindering high-resolution, continuous mapping. This study presents the first integration of canopy height models derived from airborne LiDAR data across multiple countries (~16,000 km²) with 3-meter-resolution PlanetScope satellite RGB imagery, leveraging the Depth Anything V2 monocular depth estimation framework for end-to-end training. The resulting model enables accurate and scalable canopy height retrieval without requiring stereo or LiDAR inputs. Independent validation in China (~1 km²) and the United States (~116 km²) yielded biases of 0.59 m and 0.41 m, and RMSEs of 2.54 m and 5.75 m, respectively—outperforming current global products by reducing mean absolute error by approximately 1.5 m and RMSE by about 2 m.

This study addresses the performance limitations and high computational overhead of Transformers in remote sensing pixel-level regression caused by patch-size constraints. Focusing on medium-resolution satellite imagery, we propose a tree height prediction framework based on pixel-level attention. Methodologically, a pixel-wise Transformer architecture is combined with efficient attention variants and hyperparameter tuning strategies to effectively balance accuracy and efficiency. Experimental results demonstrate that pixel-level attention significantly outperforms large-patch schemes, with the proposed model surpassing conventional methods. This research provides practical architectural design guidelines for remote sensing pixel-level tasks such as biomass estimation, achieving an optimal trade-off between prediction quality and computational resources.

Accurate estimation of forest biomass, a major carbon sink, relies heavily on tree-level traits such as height and species. Unoccupied Aerial Vehicles (UAVs) capturing high-resolution imagery from a single RGB camera offer a cost-effective and scalable approach for mapping and measuring individual trees. We introduce BIRCH-Trees, the first benchmark for individual tree height and species estimation from tree-centered UAV images, spanning three datasets: temperate forests, tropical forests, and boreal plantations. We also present DINOvTree, a unified approach using a Vision Foundation Model (VFM) backbone with task-specific heads for simultaneous height and species prediction. Through extensive evaluations on BIRCH-Trees, we compare DINOvTree against commonly used vision methods, including VFMs, as well as biological allometric equations. We find that DINOvTree achieves top overall results with accurate height predictions and competitive classification accuracy while using only 54% to 58% of the parameters of the second-best approach.

This study addresses the high cost and labor intensity of traditional forest plot surveys, which hinder scalability. The authors propose a fully automated method that reconstructs circular forest plots and extracts tree parameters using only a single pass of video captured by an ordinary smartphone. This approach uniquely integrates consumer-grade hardware with SLAM, pretrained vision models, and monocular depth estimation—eliminating the need for specialized equipment. By combining trunk instance segmentation, depth maps, and camera pose estimation, along with reference-length calibration, the system operates robustly from arbitrary starting points. Evaluated in both plantation and natural forests, the method achieves diameter-at-breast-height measurement errors of 1.51 cm (MARE 3.98%) and 2.30 cm (MARE 5.69%), respectively—matching the accuracy of conventional techniques while substantially reducing cost and operational complexity.