Tree Counts Research Articles

Tree counting of tropical tree plantations using the maximum probability spectral features of high-resolution satellite images and drones

Tree counting of tropical tree plantations using the maximum probability spectral features of high-resolution satellite images and drones

A Mathematical Model for Estimating Carbon Storage Dynamics of Forest Communities

Using the JABOWA single-tree growth model, a program was designed to estimate carbon storage dynamics in the aboveground biomass of a mixed forest community. The developed model incorporates the parameters of tree species that are common to the forests of Central Russia: pedunculate oak (Quercus robur L.), silver birch (Betula pendula Roth), common aspen (Populus tremula L.), small-leaved lime (Tilia cordata Mill.), Scots pine (Pinus sylvestris L.), Norway spruce (Picea abies (L.) H. Karst.), and fir (Abies Mill.). A differential equation for tree diameter at breast height (D) was solved. The results were compared with the forest inventory data. The amount of carbon stored in the aboveground biomass of trees was calculated following the methodology suggested by the Intergovernmental Panel on Climate Change. The dynamics of tree volume were analyzed. An analytical formula was proposed to describe the dependence of tree volume and stored carbon on tree age. The differences in the rates of tree volume growth and carbon accumulation were identified among the species studied. The analytical and numerical results on stored carbon and tree age showed a good agreement for a test plot with the known species composition and tree count, which is located within the forest part of the carbon polygon of Kazan Federal University. The formula offers an accurate estimation and prediction of carbon storage dynamics in mixed forest communities with trees varying in age and, hence, is a valuable tool for managing forestry activities. However, when predicting tree biomass growth and carbon storage dynamics, one should also consider forest site quality classes reflecting the actual growth conditions of trees. Developing a mathematical model based on forest site quality classes as a key variable would help increase the reliability of biomass growth and carbon storage predictions for forest communities. Notably, the obtained model applies to actual forest communities with known species composition and fails to account for natural regeneration. To incorporate this parameter, spatial diffusion models that describe forest regeneration in non-forest areas should be utilized.

Dynamics of stand density and self-thinning in Chinese fir plantations: theoretical insights and empirical validation.

Stand density management is essential for adaptive silviculture, thinning decisions, growth modeling, and yield prediction in forestry, particularly for plantations. Despite extensive research on self-thinning rules and the maximum size-density law, significant gaps remain in the biophysical understanding and validation of the relationships among key stand variables and parameters. This study theoretically explored and validated the relationship between maximum size-density and two key metrics: average diameter at breast height (D) and tree height (H). We used time-series data from a 30-year clear-cut, fully stocked Chinese fir plantation, a fast-growing commercial species in China, for validation. A growth balance status for fully stocked stands was proposed, wherein prior to self-thinning, the growth rate of the stand basal area (G) aligns with that of the average tree height (H), expressed as and approaching a constant slope, b1. Generalized maximum size-density and stand density index (SDI) equations were developed: and with , differing from traditional equations. Additionally, a generalized self-thinning equation, or , was introduced, indicating that in fully stocked stands, tree volume or biomass depends on both tree height and tree count. These findings advance understanding of the maximum size-density law and self-thinning boundary, providing refined tools for managing stand density in Chinese fir plantations.

Combining OBIA, CNN, and UAV imagery for automated detection and mapping of individual olive trees

The identification of individual trees is an important research topic in forestry, remote sensing and computer vision. It represents a tool for effectively and efficiently managing and maintaining forests and orchards. However, this task is not as simple as it seems; tree detection and counting can be time consuming, cost-prohibitive and accuracy-limited, especially if performed manually on a large scale.The availability of very high-resolution UAV imagery with remote sensing can make the counting process easier, faster and more precise. With the development of technology, this process can be made more automated by using intelligent algorithms such as CNN.This work presents an OBIA-CNN (Object Based Image Analysis-Convolution Neural Network) approach that combines CNNs with OBIA to automatically detect and count olive trees from Phantom4 advanced drone imagery. Initially, The CNN-based classifier was created, trained, validated, and applied to generate the Olive trees probability maps on the ortho-photo. The post-classification refinement based on OBIA was then conducted. A super-pixel segmentation and the Excess Green index were performed and a detailed accuracy analysis has been carried out to establish the suitability of the proposed method.The application to a RGB ortho-mosaic of an olive grove, in the east region of Morocco was successful using a manually elaborated training dataset of 4500 images of 24×24 pixels. Finally, the CNN detected and counted 2934 olive trees on the ortho-photo, achieving an overall accuracy of 97 % and 99 % after the OBIA refinement. The results of the proposed OBIA-CNN method were also compared with the classification results of using the Template matching technique, CNN method alone, and OBIA analysis alone to evaluate the performance of the approach. Our findings suggest the use of very high resolution images with object-based deep learning is promising for automatic detection and counting of olive trees to support the accurate and sustainable agricultural monitoring.

Measuring ornamental tree canopy attributes for precision spraying using drone technology and self-supervised segmentation

Tree canopy attributes or characteristics, such as canopy volume and density, are important parameters for calculating the precise quantity of agrochemicals required in each tree. Assessing the agrochemical needs in a nursery can help growers effectively estimate expenses, mitigate the risks of overuse or underuse, enhance resource utilization, and promote sustainable agricultural practices. This study aimed to develop an unmanned aerial vehicle (UAV)-based system to measure tree canopy attributes like tree height, tree count, canopy area and canopy volume using an image processing and self-supervised zero-shot segmentation approach. A high-resolution red-green-blue (RGB) sensor mounted on a drone captured three aerial image datasets, D1 and D2 from the first experimental plot and D3 from the second experimental plot. The acquired aerial images were stitched together and processed to generate a digital surface model (DSM) and a digital terrain model (DTM). A self-supervised zero-shot segmentation model, Segment Anything Model (SAM), was applied for tree segmentation and counting. The tree canopy area was calculated by segmenting individual trees and applying a conversion factor to convert pixelwise areas into square meters. Height calculation was done using elevation data of each pixel from DSM-DTM imagery within SAM-derived bounding box coordinates, and the highest pixel was considered as the height of the tree. The drone-calculated heights of 24 randomly selected trees were compared with manually measured heights in all datasets. The results showed an average absolute error of 2.43% in D1, 7.09% in D2 and 12.58% in D3. The Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) were noted to be 0.09 m and 0.23 m in D1, 0.13 m and 0.25 m in D2, and 0.16 m and 0.21 m in D3, respectively. In D3, manual area and volume calculations were performed for validation. The RMSE and MAE for area calculation were calculated as 0.18 m2 and 0.15 m2, respectively, and the RMSE and MAE for volume measurements were calculated as 0.33 m3 and 0.26 m3 respectively, which indicated the level of agreement between manual and drone measurements. Using canopy volume information, calculated by summing pixel heights within each tree’s bounding box and multiplying by the ground sample distance (GSD) of 0.025 m per pixel, the average quantity of agrochemicals needed for a Maple tree was estimated at 0.20 L. These results underscore the high potential of this method for accurate canopy characteristic calculations and precision spray applications in the ornamental industry.

The Impact of Tree Density on Automated Oil Palm Tree Counting Accuracy

The rapid development of oil palm plantations in Kualuh Leidong Subdistrict, Labuhanbatu Utara District, North Sumatra Province has led to an increased need for effective and efficient monitoring and supervision of oil palm trees. One method that supports such monitoring and supervision is the use of an automatic counting method using orthophoto data. This orthophoto data was used for automatic tree counting using a deep learning method with the Faster R-CNN algorithm. The study considered two planting patterns: regular planting patterns with spacing of 4 to 9 meters, and random planting patterns with varying spacing. Data processing involved an epoch value of 80 and a batch size value of 4. The accuracy of the automatic oil palm tree counting was evaluated based on the density of the spacing between trees with reference to the ground truth. The findings indicated that the deep learning Faster R-CNN algorithm achieved higher accuracy in automatic calculations for regular planting patterns.

Hyperfiniteness for group actions on trees

We identify natural conditions for a countable group acting on a countable tree which imply that the orbit equivalence relation of the induced action on the Gromov boundary is Borel hyperfinite. Examples of this condition include acylindrical actions. We also identify a natural weakening of the aforementioned conditions that implies measure hyperfiniteness of the boundary action. We then document examples of group actions on trees whose boundary action is not hyperfinite.

AdaTreeFormer: Few shot domain adaptation for tree counting from a single high-resolution image

The process of estimating and counting tree density using only a single aerial or satellite image is a difficult task in the fields of photogrammetry and remote sensing. However, it plays a crucial role in the management of forests. The huge variety of trees in varied topography severely hinders tree counting models to perform well. The purpose of this paper is to propose a framework that is learnt from the source domain with sufficient labeled trees and is adapted to the target domain with only a limited number of labeled trees. Our method, termed as AdaTreeFormer, contains one shared encoder with a hierarchical feature extraction scheme to extract robust features from the source and target domains. It also consists of three subnets: two for extracting self-domain attention maps from source and target domains respectively and one for extracting cross-domain attention maps. For the latter, an attention-to-adapt mechanism is introduced to distill relevant information from different domains while generating tree density maps; a hierarchical cross-domain feature alignment scheme is proposed that progressively aligns the features from the source and target domains. We also adopt adversarial learning into the framework to further reduce the gap between source and target domains. Our AdaTreeFormer is evaluated on six designed domain adaptation tasks using three tree counting datasets, i.e. Jiangsu, Yosemite, and London. Experimental results show that AdaTreeFormer significantly surpasses the state of the art, e.g. in the cross domain from the Yosemite to Jiangsu dataset, it achieves a reduction of 15.9 points in terms of the absolute counting errors and an increase of 10.8% in the accuracy of the detected trees’ locations. The codes and datasets are available at https://github.com/HAAClassic/AdaTreeFormer.

Optimizing Oil Palm Tree Inventory Management with Advanced Drone-Based Image Recognition Techniques

Oil palm tree plantations are a crucial source of vegetable oil for various industries, playing a significant role in the global economy. One key aspect of plantation management is accurate inventory management of oil palm trees, which involves the detection and counting of individual trees. Traditional inventory management methods for oil palm trees rely on ground-based manual measurements, which are time-consuming, labor-intensive, and prone to errors. However, accurate and efficient detection and counting of oil palm trees from drone images remain challenging due to the complex and variable nature of the plantation environment. Drone-based remote sensing has emerged as a promising alternative for inventory management in recent years. In this study, a novel approach to enhance the accuracy and efficiency of oil palm tree detection and counting using advanced drone-based image recognition techniques is proposed. The research discusses a novel image recognition technique that uses a custom GLCM, Haar Wavelet, and template matching to detect and count oil palm trees from drone images. The proposed approach outperformed traditional machine learning techniques and achieved a high accuracy of 83.75% and 86.9% in detecting individual oil palm trees in Jeli and Keratong, respectively. Haar Wavelet proves this algorithm achieves the highest overall accuracy. Additionally, eight statistical parameters can manipulate the GLCM, while offset parameters can boost accuracy for various applications or methods. The study highlights the potential of advanced drone-based image recognition techniques for optimizing plantation management and contributing to sustainable production practices.

Estimation of Oil Palm Tree Quantity Using Oil Palm Ecognition and SAGA GIS Based on UAV Imagery (Case Study: Dumai, Riau)

Abstract Oil palm is a plantation commodity in Indonesia that drives the growth of the national economy. The growth of oil palm is rapidly expanding every year, thus requiring efficient land monitoring strategies. Tree counting is one technique used to monitor current conditions in vast plantation areas. Utilizing UAV technology in agricultural mapping is the best option as a tool for capturing high-resolution aerial photos. The objective of this research is to automatically count the number of oil palm trees by comparing two methods and algorithms. GEOBIA (Geography-Based Image Analysis) and Deep Learning classification are methods used for automatic object counting. In this study, high-resolution aerial photo data was utilized, making object detection much more feasible. The research applied the Template Matching and Watershed Segmentation algorithm models in a sample area of 32.5 hectares divided into two blocks: Unknown-2 and E-1. It can be concluded that both methods, with their respective algorithm models, are sufficiently relevant for use in the automatic counting process of oil palm trees. This is evidenced by the accuracy test error values, which are not more than 15%, indicating that the counted oil palm tree results can be used for further data analysis.

Examination of Drone Usage in Estimating Hardwood Plantations Structural Metrics

Planting hardwood trees on retired marginal agricultural land is one of the main strategies used to restore forested wetlands. Evaluating effectiveness of wetland restoration requires efficient monitoring to evaluate recovery trajectories and desired conditions. Recent advancements in unmanned aerial system (UAS) technologies have prompted wide-scale adoption of UAS platforms in providing a range of ecological data. In this study, we examined the use of UAS Structure from Motion (SfM) derived point clouds in estimating tree density, canopy height, and percent canopy cover for bottomland hardwood plantations within four wetland reserve easements. Using a local maxima approach for individual tree detection produced plantation level estimates with mean absolute errors of 150 trees per hectare, 0.5 m, and 18.4% for tree density, canopy height, and percent canopy cover, respectively. At the plot level, UAS-derived tree counts (r = 0.53, p < 0.01) and canopy height (r = 0.57, p < 0.01) were significantly correlated with ground-based estimates. We demonstrate that UAS-SfM is a viable method of assessing bottomland hardwood plantations for applications that require precision levels congruent with the mean absolute errors reported here. The accuracy of tree density estimates was reliant upon specific local maxima window parameters relative to stand conditions. Therefore, acquisition of leaf-off and leaf-on imagery may allow for better individual tree detection and subsequently more accurate tree density and other structural attributes.

The Number of Topological Types of Trees

Two graphs are of the same topological type if they can be mutually embedded into each other topologically. We show that there are exactly ℵ1\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\aleph _1$$\\end{document} distinct topological types of countable trees. In general, for any infinite cardinal κ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\kappa $$\\end{document} there are exactly κ+\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\kappa ^+$$\\end{document} distinct topological types of trees of size κ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\kappa $$\\end{document}. This solves a problem of van der Holst from 2005.

Provisioning services decline for both people and Critically Endangered wildlife in a rainforest transformation landscape

Abstract The loss and degradation of forests and other ecosystems worldwide threaten both global biodiversity and the livelihoods of people who use natural resources. Understanding how natural resource use impacts landscape provisioning services for both people and wildlife is thus critical for designing comprehensive resource management strategies. We used data from community focus groups, botanical plots and an inventory of plant species consumed by the Critically Endangered red‐ruffed lemur (Varecia rubra) to assess the availability of key provisioning services for people and endemic wildlife on the Masoala Peninsula, a rainforest transformation landscape, in northeastern Madagascar (Masoala National Park and 13 surrounding communities). We constructed Poisson regression mixed models to evaluate the impact of community factors (i.e. community population size, plot distance to community) and changes over time on the count and species richness of timber trees, medicinal plants and red‐ruffed lemur food trees within botanical plots. Over three‐quarters of all plant species could be used for at least one purpose by local communities (n = 238 species). Of the 59 V. rubra food tree species, only 15% had no reported human use. Timber and ruffed lemur food tree availability declined both with community population size and time and were predicted to be lower outside of Masoala National Park. In contrast, medicinal plant availability was not strongly predicted by any tested factors. Provisioning service availability also differed strongly across sites, suggesting that additional, untested proxies of human pressure likely also have an effect. Our results highlight the importance of evaluating natural resource availability from a community‐based perspective and by resource purpose to inform forest landscape restoration efforts that can support both people and wildlife. Read the free Plain Language Summary for this article on the Journal blog.

Application of Image Analytics for Tree Enumeration for Diversion of Forest Land

Accurate tree enumeration is essential for responsible forest land diversion in development projects. Conventional manual surveys are slow, costly, and prone to errors. This paper introduces a cutting-edge image analytics solution that leverages satellite imagery and aerial photos to automate tree counting. The primary objective is to develop a robust system that identifies and categorizes trees by crown size, and environmental conditions. Advanced computer vision algorithms are integrated with machine learning models to analyze the imagery. Rigorous validation processes, including comparisons with ground-truth data from manual surveys, ensure high accuracy and reliability. The results are impressive. This solution significantly accelerates tree enumeration, eliminating resource-intensive manual efforts. It consistently demonstrates precision with minimal false positives and negatives. Moreover, it categorizes trees efficiently by species, offering a comprehensive view of the forested area. This project significance lies in its contribution to responsible and sustainable land development practices. By automating tree enumeration, it equips stakeholders with timely, precise data for informed decisions about land usage, conservation, and environmental impact assessments. The solution strikes a balance between development and ecological preservation, optimizing resource allocation while minimizing environmental impact in forested regions. In conclusion, this innovative image analytics solution revolutionizes forest land diversion, enabling efficient and ecologically conscious decision-making. It addresses the critical need for accurate tree enumeration in the face of developmental challenges, fostering responsible land use and environmental stewardship.

A comparative study on the efficacy of dual-pol and full-pol ASAR data in radiative transfer modeling for forest above-ground biomass estimation

ABSTRACT In this study, a comprehensive evaluation of the potential of dual-polarization and full-polarization L- and S-band airborne SAR (LS-ASAR) data (ISRO’s dual-frequency airborne SAR mounted on NASA’s Gulfstream III aircraft) in forest above-ground biomass (AGB) estimation has been carried out. The study area comprises the temperate forests of the Lenoir landing site in Southwest Alabama, United States of America. A trunk scattering model based on Vector Radiative Transfer (VRT) theory is used in this study to estimate the AGB. Full-polarization ASAR data inversion retrieves three critical biophysical parameters: tree height, diameter at breast height (DBH), and tree count, whereas with dual-polarization ASAR data, the model is limited to retrieving only two parameters (tree height and DBH). These parameters are then employed in a suitable allometric equation to estimate the AGB over the study area. Validation of the retrieval results is performed through comparison with ground-truth measurements taken from the study site. The findings demonstrate that the inversion of L-band full-polarized data yields the highest correlation (R2 = 0.87, RMSE = 20.37 t/ha), revealing high accuracy in AGB estimation. The L-band dual-polarized data shows a high correlation (R2 = 0.83, RMSE = 30.66 t/ha), indicating reasonable accuracy. However, the S-band data, both full and dual-polarized, reveals weaker correlations with higher RMSE values (R2 = 0.59, RMSE = 55.07 t/ha, and R2 = 0.49, RMSE = 58.31 t/ha, respectively), suggesting a lower degree of reliability. This study emphasizes the pre-eminence of L-band fully polarized data inversion for reliable and accurate AGB estimation while also showing the utility of relatively ubiquitous dual-pol data (L-band) to achieve reasonable accuracy for the same.

An Open Benchmark Dataset for Forest Characterization from Sentinel-1 and -2 Time Series

Earth observation satellites offer vast opportunities for quantifying landscapes and regional land cover composition and changes. The integration of artificial intelligence in remote sensing is essential for monitoring significant land cover types like forests, demanding a substantial volume of labeled data for effective AI model development and validation. The Wald5Dplus project introduces a distinctive open benchmark dataset for mid-European forests, labeling Sentinel-1/2 time series using data from airborne laser scanning and multi-spectral imagery. The freely accessible satellite images are fused in polarimetric, spectral, and temporal domains, resulting in analysis-ready data cubes with 512 channels per year on a 10 m UTM grid. The dataset encompasses labels, including tree count, crown area, tree types (deciduous, coniferous, dead), mean crown volume, base height, tree height, and forested area proportion per pixel. The labels are based on an individual tree characterization from high-resolution airborne LiDAR data using a specialized segmentation algorithm. Covering three test sites (Bavarian Forest National Park, Steigerwald, and Kranzberg Forest) and encompassing around six million trees, it generates over two million labeled samples. Comprehensive validation, including metrics like mean absolute error, median deviation, and standard deviation, in the random forest regression confirms the high quality of this dataset, which is made freely available.

Early-Stage Pine Wilt Disease Detection via Multi-Feature Fusion in UAV Imagery

Pine wilt disease (PWD) is a highly contagious and devastating forest disease. The timely detection of pine trees infected with PWD in the early stage is of great significance to effectively control the spread of PWD and protect forest resources. However, in the spatial domain, the features of early-stage PWD are not distinctly evident, leading to numerous missed detections and false positives when directly using spatial-domain images. However, we found that frequency domain information can more clearly express the characteristics of early-stage PWD. In this paper, we propose a detection method based on deep learning for early-stage PWD by comprehensively utilizing the features in the frequency domain and the spatial domain. An attention mechanism is introduced to further enhance the frequency domain features. Employing two deformable convolutions to fuse the features in both domains, we aim to fully capture semantic and spatial information. To substantiate the proposed method, this study employs UAVs to capture images of early-stage pine trees infected with PWD at Dahuofang Experimental Forest in Fushun, Liaoning Province. A dataset of early infected pine trees affected by PWD is curated to facilitate future research on the detection of early-stage infestations in pine trees. The results on the early-stage PWD dataset indicate that, compared to Faster R-CNN, DETR and YOLOv5, the best-performing method improves the average precision (AP) by 17.7%, 6.2% and 6.0%, and the F1 scores by 14.6%, 3.9% and 5.0%, respectively. The study provides technical support for early-stage PWD tree counting and localization in the field in forest areas and lays the foundation for the early control of pine wood nematode disease.

Coupling an Autonomous UAV With a ML Framework for Sustainable Environmental Monitoring and Remote Sensing

Many countries face problems in monitoring plant problems and monitoring large environments, as until now, there is no accurate means of aerial monitoring through which concerned parties can benefit from watersheds, monitor large agricultural areas, and make wise environmental decisions regarding them. This paper describes a pioneering approach to develop smart agriculture using multimission drones equipped with dual cognitive modules (brains) that are powered by a machine learning (ML) framework. The first brain uses deep reinforcement learning (DRL) principles to enable autonomous flight, allowing drones to navigate complex agricultural terrain with agility and flexibility. The second brain is responsible for precise and crucial agricultural tasks: counting trees, detecting water locations, and observing and analyzing plants using the Faster R‐CNN algorithm. The system is linked with the ground station for control and command, as well as it includes an Internet of Things (IoT) infrastructure equipped with sensors that collect soil parameters, which then get sent via 5G Wi‐Fi. The dual architecture of the drones, combined with the ground‐level IoT system, creates a comprehensive framework that not only enhances agricultural technologies but also aligns with environmental conservation goals, embodying a paradigm shift towards a greener and more sustainable future. Obtained results show reasonable results with an accuracy rate of 98%.

Template matching technique for smart tree detection and counting with UAV imagery

The identification of individual trees is an important research topic in forestry, remote sensing, and computer vision. It is a requirement in forest management and monitoring because it provides key forest inventory information, vegetation distribution mapping, vegetation density estimation, change monitoring, and species classification.However, counting trees can be challenging due to the crowded environment, time-consumption, and expensive operation. Remote Sensing methods such as UAV imagery and the development of efficient algorithms can be adapted to estimate and detect individual tree counts in orchards. This paper aims to use the template matching technique to automatically detect olive trees from high resolution drone imagery in the eastern part of Morocco. The algorithm successfully detected and counted 2719 olive trees with a difference of less than 233 trees with manual detection. The results of detecting and counting the individual olive trees were evaluated using several parameters: an Fscore of 94%, with a recall of 92% and a precision of 98%, which are satisfactory.

Tree-CRowNN: A Network for Estimating Forest Stand Density from VHR Aerial Imagery

Estimating the number of trees within a forest stand, i.e., the forest stand density (FSD), is challenging at large scales. Recently, researchers have turned to a combination of remote sensing and machine learning techniques to derive these estimates. However, in most cases, the developed models rely heavily upon additional data such as LiDAR-based elevations or multispectral information and are mostly applied to managed environments rather than natural/mixed forests. Furthermore, they often require the time-consuming manual digitization or masking of target features, or an annotation using a bounding box rather than a simple point annotation. Here, we introduce the Tree Convolutional Row Neural Network (Tree-CRowNN), an alternative model for tree counting inspired by Multiple-Column Neural Network architecture to estimate the FSD over 12.8 m × 12.8 m plots from high-resolution RGB aerial imagery. Our model predicts the FSD with very high accuracy (MAE: ±2.1 stems/12.8 m2, RMSE: 3.0) over a range of forest conditions and shows promise in linking to Sentinel-2 imagery for broad-scale mapping (R2: 0.43, RMSE: 3.9 stems/12.8 m2). We believe that the satellite imagery linkage will be strengthened with future efforts, and transfer learning will enable the Tree-CRowNN model to predict the FSD accurately in other ecozones.