Forests are vital ecosystems that play a crucial role in sustaining life on Earth and supporting human well-being.Traditional forest mapping and monitoring methods are often costly and limited in scope,necessitating t...Forests are vital ecosystems that play a crucial role in sustaining life on Earth and supporting human well-being.Traditional forest mapping and monitoring methods are often costly and limited in scope,necessitating the adoption of advanced,automated approaches for improved forest conservation and management.This study explores the application of deep learning-based object detection techniques for individual tree detection in RGB satellite imagery.A dataset of 3157 images was collected and divided into training(2528),validation(495),and testing(134)sets.To enhance model robustness and generalization,data augmentation was applied to the training part of the dataset.Various YOLO-based models,including YOLOv8,YOLOv9,YOLOv10,YOLOv11,and YOLOv12,were evaluated using different hyperparameters and optimization techniques,such as stochastic gradient descent(SGD)and auto-optimization.These models were assessed in terms of detection accuracy and the number of detected trees.The highest-performing model,YOLOv12m,achieved a mean average precision(mAP@50)of 0.908,mAP@50:95 of 0.581,recall of 0.851,precision of 0.852,and an F1-score of 0.847.The results demonstrate that YOLO-based object detection offers a highly efficient,scalable,and accurate solution for individual tree detection in satellite imagery,facilitating improved forest inventory,monitoring,and ecosystem management.This study underscores the potential of AI-driven tree detection to enhance environmental sustainability and support data-driven decision-making in forestry.展开更多
Single-tree detection is one of the main research topics in quantifying the structural properties of forests. Drone Li DAR systems and terrestrial laser scanning systems produce high-density point clouds that offer gr...Single-tree detection is one of the main research topics in quantifying the structural properties of forests. Drone Li DAR systems and terrestrial laser scanning systems produce high-density point clouds that offer great promise for forest inventories in limited areas. However, most studies have focused on the upper canopy layer and neglected the lower forest structure. This paper describes an innovative tree detection method using drone Li DAR data from a new perspective of the under-canopy structure. This method relies on trunk point clouds, with undercanopy sections split into heights ranging from 1 to 7 m, which were processed and compared, to determine a suitable height threshold to detect trees. The method was tested in a dense cedar plantation forest in the Aichi Prefecture, Japan, which has a stem density of 1140 stems·ha^(-1) and an average tree age of 42 years. Dense point cloud data were generated from the drone Li DAR system and terrestrial laser scanning with an average point density of 5000 and 6500 points·m^(-2), respectively. Tree detection was achieved by drawing point-cloud section projections of tree trunks at different heights and calculating the center coordinates. The results show that this trunk-section-based method significantly reduces the difficulty of tree detection in dense plantation forests with high accuracy(F1-Score=0.9395). This method can be extended to different forest scenarios or conditions by changing section parameters.展开更多
To avoid colliding with trees during its operation,a lawn mower robot must detect the trees.Existing tree detection methods suffer from low detection accuracy(missed detection)and the lack of a lightweight model.In th...To avoid colliding with trees during its operation,a lawn mower robot must detect the trees.Existing tree detection methods suffer from low detection accuracy(missed detection)and the lack of a lightweight model.In this study,a dataset of trees was constructed on the basis of a real lawn environment.According to the theory of channel incremental depthwise convolution and residual suppression,the Embedded-A module is proposed,which expands the depth of the feature map twice to form a residual structure to improve the lightweight degree of the model.According to residual fusion theory,the Embedded-B module is proposed,which improves the accuracy of feature-map downsampling by depthwise convolution and pooling fusion.The Embedded YOLO object detection network is formed by stacking the embedded modules and the fusion of feature maps of different resolutions.Experimental results on the testing set show that the Embedded YOLO tree detection algorithm has 84.17%and 69.91%average precision values respectively for trunk and spherical tree,and 77.04% mean average precision value.The number of convolution parameters is 1.78×10^(6),and the calculation amount is 3.85 billion float operations per second.The size of weight file is 7.11MB,and the detection speed can reach 179 frame/s.This study provides a theoretical basis for the lightweight application of the object detection algorithm based on deep learning for lawn mower robots.展开更多
A difficult problem in forestry is tree inventory.In this study, a GoProHero attached to a small unmanned aerial vehicle was used to capture images of a small area covered by pinus pinea trees. Then, a digital surface...A difficult problem in forestry is tree inventory.In this study, a GoProHero attached to a small unmanned aerial vehicle was used to capture images of a small area covered by pinus pinea trees. Then, a digital surface model was generated with image matching. The elevation model representing the terrain surface, a ‘digital terrain model’,was extracted from the digital surface model using morphological filtering. Individual trees were extracted by analyzing elevation flow on the digital elevation model because the elevation reached the highest value on the tree peaks compared to the neighborhood elevation pixels. The quality of the results was assessed by comparison with reference data for correctness of the estimated number of trees. The tree heights were calculated and evaluated with ground truth dataset. The results showed 80% correctness and 90% completeness.展开更多
Trees are an integral part of the forestry ecosystem.In forestry work,the precise acquisition of tree morphological parameters and attributes is affected by complex illumination and tree morphology.In order to minimiz...Trees are an integral part of the forestry ecosystem.In forestry work,the precise acquisition of tree morphological parameters and attributes is affected by complex illumination and tree morphology.In order to minimize a series of inestimable problems,such as yield reduction,ecological damage,and destruction,caused by inaccurate acquisition of tree location information,this paper proposes a ground tree detection method GMOSTNet.Based on the four types of tree species in the GMOST dataset and Faster R-CNN,it extracted the features of the trees,generate candidate regions,classification,and other operations.By reducing the influence of illumination and occlusion factors during experimentation,more detailed information of the input image was obtained.Meanwhile,regarding false detections caused by inappropriate approximations,the deviation and proximity of the proposal were adjusted.The experimental results showed that the AP value of the four tree species is improved after using GMOSTNet,and the overall accuracy increases from the original 87.25%to 93.25%.展开更多
Rapid urbanization has caused significant changes along the urban-rural gradient,leading to a variety of landscapes that are mainly shaped by human activities.This dynamic interplay also influences the distribution an...Rapid urbanization has caused significant changes along the urban-rural gradient,leading to a variety of landscapes that are mainly shaped by human activities.This dynamic interplay also influences the distribution and characteristics of trees outside forests(TOF).Understanding the pattern of these trees will support informed decision-making in urban planning,in conservation strategies,and altogether in sustainable land management practices in the urban context.In this study,we employed a deep learning-based object detection model and high resolution satellite imagery to identify 1.3 million trees with bounding boxes within a 250 km^(2)research transect spanning the urban-rural gradient of Bengaluru,a megacity in Southern India.Additionally,we developed an allometric equation to estimate diameter at breast height(DBH)from the tree crown diameter(CD)derived from the detected bounding boxes.Our study focused on analyzing variations in tree density and tree size along this gradient.The findings revealed distinct patterns:the urban domain displayed larger tree crown diameters(mean:8.87 m)and DBH(mean:43.78 cm)but having relatively low tree density(32 trees per hectare).Furthermore,with increasing distance from the city center,tree density increased,while the mean tree crown diameter and mean tree basal area decreased,showing clear differences of tree density and size between the urban and rural domains in Bengaluru.This study offers an efficient methodology that helps generating instructive insights into the dynamics of TOF along the urban-rural gradient.This may inform urban planning and management strategies for enhancing green infrastructure and biodiversity conservation in rapidly urbanizing cities like Bengaluru.展开更多
Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facili...Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facilitating safe and flexible navigation in dense forests.However,their limited onboard computational power makes real-time,image-based tree trunk segmentation challenging,emphasizing the urgent need for lightweight and efficient segmentation models.In this study,we present RT-Trunk,a model specifically designed for real-time tree trunk instance segmentation in complex forest environments.To ensure real-time performance,we selected SparseInst as the base framework.We incorporated ConvNeXt-T as the backbone to enhance feature extraction for tree trunks,thereby improving segmentation accuracy.We further integrate the lightweight convolutional block attention module(CBAM),enabling the model to focus on tree trunk features while suppressing irrelevant information,which leads to additional gains in segmentation accuracy.To enable RT-Trunk to operate effectively under diverse complex forest environments,we constructed a comprehensive dataset for training and testing by combining self-collected data with multiple public datasets covering different locations,seasons,weather conditions,tree species,and levels of forest clutter.Com-pared with the other tree trunk segmentation methods,the RT-Trunk method achieved an average precision of 91.4%and the fastest inference speed of 32.9 frames per second.Overall,the proposed RT-Trunk provides superior trunk segmentation performance that balances speed and accu-racy,making it a promising solution for supporting under-canopy UAVs in the autonomous extraction of standing tree stem attributes.The code for this work is available at https://github.com/NEFU CVRG/RT Trunk.展开更多
Oil palm tree monitoring is essential to track productivity and prevent diseases,but manual methods are labor-intensive,cost-prohibitive,and error-prone.While existing remote sensing data and deep learning methods off...Oil palm tree monitoring is essential to track productivity and prevent diseases,but manual methods are labor-intensive,cost-prohibitive,and error-prone.While existing remote sensing data and deep learning methods offer effi cient alternatives for monitor-ing,achieving a highly accurate tree condition classification remains a challenge.Therefore,this study classifies healthy,yellow,small,and dead oil palm trees using the YOLOv8 model.A publicly available multi-class oil palm tree dataset,after extensive correction of identified labeling errors,is used to train the model.The model's performance is compared with state-of-the-art object detectors,and a prototype web application is developed to test the model on adverse unseen image scenes.The YOLOv8 model achieves up to 99.7%F1-score and 99.3%mAP across all classes of the corrected dataset.It outperforms other object detectors and produces similar scores as the recent YOLOv10-large model.Further validation on unseen images from the developed prototype results in 76.0%F1-score and 77.9%mAP across all classes.Finally,implications empha-size the role of YoLOv8 in handling class imbalance resulting from underrepresented classes in the dataset.The experimental findings,practical demonstration,and implications presented in this paper offer robust and reliable monitoring of oil palm trees with innova-tions in precision agriculture.展开更多
Purpose-Taking into consideration the current human need for agricultural produce such as rice that requires water for growth,the optimal consumption of this valuable liquid is important.Unfortunately,the traditional ...Purpose-Taking into consideration the current human need for agricultural produce such as rice that requires water for growth,the optimal consumption of this valuable liquid is important.Unfortunately,the traditional use of water by humans for agricultural purposes contradicts the concept of optimal consumption.Therefore,designing and implementing a mechanized irrigation system is of the highest importance.This system includes hardware equipment such as liquid altimeter sensors,valves and pumps which have a failure phenomenon as an integral part,causing faults in the system.Naturally,these faults occur at probable time intervals,and the probability function with exponential distribution is used to simulate this interval.Thus,before the implementation of such high-cost systems,its evaluation is essential during the design phase.Design/methodology/approach-The proposed approach included two main steps:offline and online.The offline phase included the simulation of the studied system(i.e.the irrigation system of paddy fields)and the acquisition of a data set for training machine learning algorithms such as decision trees to detect,locate(classification)and evaluate faults.In the online phase,C5.0 decision trees trained in the offline phase were used on a stream of data generated by the system.Findings-The proposed approach is a comprehensive online component-oriented method,which is a combination of supervisedmachine learning methods to investigate system faults.Each of thesemethods is considered a component determined by the dimensions and complexity of the case study(to discover,classify and evaluate fault tolerance).These components are placed together in the form of a process framework so that the appropriate method for each component is obtained based on comparison with other machine learning methods.As a result,depending on the conditions under study,the most efficient method is selected in the components.Before the system implementation phase,its reliability is checked by evaluating the predicted faults(in the system design phase).Therefore,this approach avoids the construction of a high-risk system.Compared to existing methods,the proposed approach is more comprehensive and has greater flexibility.Research limitations/implications-By expanding the dimensions of the problem,the model verification space grows exponentially using automata.Originality/value-Unlike the existing methods that only examine one or two aspects of fault analysis such as fault detection,classification and fault-tolerance evaluation,this paper proposes a comprehensive processoriented approach that investigates all three aspects of fault analysis concurrently.展开更多
基金funding from Horizon Europe Framework Programme(HORIZON),call Teaming for Excellence(HORIZON-WIDERA-2022-ACCESS-01-two-stage)-Creation of the centre of excellence in smart forestry“Forest 4.0”No.101059985funded by the EuropeanUnion under the project FOREST 4.0-“Ekscelencijos centras tvariai miško bioekonomikai vystyti”No.10-042-P-0002.
文摘Forests are vital ecosystems that play a crucial role in sustaining life on Earth and supporting human well-being.Traditional forest mapping and monitoring methods are often costly and limited in scope,necessitating the adoption of advanced,automated approaches for improved forest conservation and management.This study explores the application of deep learning-based object detection techniques for individual tree detection in RGB satellite imagery.A dataset of 3157 images was collected and divided into training(2528),validation(495),and testing(134)sets.To enhance model robustness and generalization,data augmentation was applied to the training part of the dataset.Various YOLO-based models,including YOLOv8,YOLOv9,YOLOv10,YOLOv11,and YOLOv12,were evaluated using different hyperparameters and optimization techniques,such as stochastic gradient descent(SGD)and auto-optimization.These models were assessed in terms of detection accuracy and the number of detected trees.The highest-performing model,YOLOv12m,achieved a mean average precision(mAP@50)of 0.908,mAP@50:95 of 0.581,recall of 0.851,precision of 0.852,and an F1-score of 0.847.The results demonstrate that YOLO-based object detection offers a highly efficient,scalable,and accurate solution for individual tree detection in satellite imagery,facilitating improved forest inventory,monitoring,and ecosystem management.This study underscores the potential of AI-driven tree detection to enhance environmental sustainability and support data-driven decision-making in forestry.
基金funded by KAKENHI Number 16H02556 of the Cabinet Office,Government of Japan,the Cross-ministerial Strategic Innovation Promotion Program(SIP),“Enhancement of Societal Resiliency Against Natural Disasters”Funding was provided by the Japan Science and Technology Agency(JST)as part of the Belmont ForumThis work was supported by JST SPRING,Grant Number JPMJSP2124。
文摘Single-tree detection is one of the main research topics in quantifying the structural properties of forests. Drone Li DAR systems and terrestrial laser scanning systems produce high-density point clouds that offer great promise for forest inventories in limited areas. However, most studies have focused on the upper canopy layer and neglected the lower forest structure. This paper describes an innovative tree detection method using drone Li DAR data from a new perspective of the under-canopy structure. This method relies on trunk point clouds, with undercanopy sections split into heights ranging from 1 to 7 m, which were processed and compared, to determine a suitable height threshold to detect trees. The method was tested in a dense cedar plantation forest in the Aichi Prefecture, Japan, which has a stem density of 1140 stems·ha^(-1) and an average tree age of 42 years. Dense point cloud data were generated from the drone Li DAR system and terrestrial laser scanning with an average point density of 5000 and 6500 points·m^(-2), respectively. Tree detection was achieved by drawing point-cloud section projections of tree trunks at different heights and calculating the center coordinates. The results show that this trunk-section-based method significantly reduces the difficulty of tree detection in dense plantation forests with high accuracy(F1-Score=0.9395). This method can be extended to different forest scenarios or conditions by changing section parameters.
基金the National Natural Science Foundation of China (No.51275223)。
文摘To avoid colliding with trees during its operation,a lawn mower robot must detect the trees.Existing tree detection methods suffer from low detection accuracy(missed detection)and the lack of a lightweight model.In this study,a dataset of trees was constructed on the basis of a real lawn environment.According to the theory of channel incremental depthwise convolution and residual suppression,the Embedded-A module is proposed,which expands the depth of the feature map twice to form a residual structure to improve the lightweight degree of the model.According to residual fusion theory,the Embedded-B module is proposed,which improves the accuracy of feature-map downsampling by depthwise convolution and pooling fusion.The Embedded YOLO object detection network is formed by stacking the embedded modules and the fusion of feature maps of different resolutions.Experimental results on the testing set show that the Embedded YOLO tree detection algorithm has 84.17%and 69.91%average precision values respectively for trunk and spherical tree,and 77.04% mean average precision value.The number of convolution parameters is 1.78×10^(6),and the calculation amount is 3.85 billion float operations per second.The size of weight file is 7.11MB,and the detection speed can reach 179 frame/s.This study provides a theoretical basis for the lightweight application of the object detection algorithm based on deep learning for lawn mower robots.
基金financially supported by the scientific research projects coordination unit of Akdeniz University,Project No.FBA-2015-446
文摘A difficult problem in forestry is tree inventory.In this study, a GoProHero attached to a small unmanned aerial vehicle was used to capture images of a small area covered by pinus pinea trees. Then, a digital surface model was generated with image matching. The elevation model representing the terrain surface, a ‘digital terrain model’,was extracted from the digital surface model using morphological filtering. Individual trees were extracted by analyzing elevation flow on the digital elevation model because the elevation reached the highest value on the tree peaks compared to the neighborhood elevation pixels. The quality of the results was assessed by comparison with reference data for correctness of the estimated number of trees. The tree heights were calculated and evaluated with ground truth dataset. The results showed 80% correctness and 90% completeness.
基金National Natural Science Foundation of China(U1809208).
文摘Trees are an integral part of the forestry ecosystem.In forestry work,the precise acquisition of tree morphological parameters and attributes is affected by complex illumination and tree morphology.In order to minimize a series of inestimable problems,such as yield reduction,ecological damage,and destruction,caused by inaccurate acquisition of tree location information,this paper proposes a ground tree detection method GMOSTNet.Based on the four types of tree species in the GMOST dataset and Faster R-CNN,it extracted the features of the trees,generate candidate regions,classification,and other operations.By reducing the influence of illumination and occlusion factors during experimentation,more detailed information of the input image was obtained.Meanwhile,regarding false detections caused by inappropriate approximations,the deviation and proximity of the proposal were adjusted.The experimental results showed that the AP value of the four tree species is improved after using GMOSTNet,and the overall accuracy increases from the original 87.25%to 93.25%.
基金financial support provided by the German Research Foundation,DFG,through grant number KL894/23-2 and NO 1444/1-2 as part of the Research Unit FOR2432/2the China Scholarship Council(CSC)that supports the first author with a Ph D scholarshipsupport provided by Indian partners at the Institute of Wood Science and Technology(IWST),Bengaluru。
文摘Rapid urbanization has caused significant changes along the urban-rural gradient,leading to a variety of landscapes that are mainly shaped by human activities.This dynamic interplay also influences the distribution and characteristics of trees outside forests(TOF).Understanding the pattern of these trees will support informed decision-making in urban planning,in conservation strategies,and altogether in sustainable land management practices in the urban context.In this study,we employed a deep learning-based object detection model and high resolution satellite imagery to identify 1.3 million trees with bounding boxes within a 250 km^(2)research transect spanning the urban-rural gradient of Bengaluru,a megacity in Southern India.Additionally,we developed an allometric equation to estimate diameter at breast height(DBH)from the tree crown diameter(CD)derived from the detected bounding boxes.Our study focused on analyzing variations in tree density and tree size along this gradient.The findings revealed distinct patterns:the urban domain displayed larger tree crown diameters(mean:8.87 m)and DBH(mean:43.78 cm)but having relatively low tree density(32 trees per hectare).Furthermore,with increasing distance from the city center,tree density increased,while the mean tree crown diameter and mean tree basal area decreased,showing clear differences of tree density and size between the urban and rural domains in Bengaluru.This study offers an efficient methodology that helps generating instructive insights into the dynamics of TOF along the urban-rural gradient.This may inform urban planning and management strategies for enhancing green infrastructure and biodiversity conservation in rapidly urbanizing cities like Bengaluru.
基金supported in part by the National Natural Science Foundation of China(No.31470714 and 61701105).
文摘Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facilitating safe and flexible navigation in dense forests.However,their limited onboard computational power makes real-time,image-based tree trunk segmentation challenging,emphasizing the urgent need for lightweight and efficient segmentation models.In this study,we present RT-Trunk,a model specifically designed for real-time tree trunk instance segmentation in complex forest environments.To ensure real-time performance,we selected SparseInst as the base framework.We incorporated ConvNeXt-T as the backbone to enhance feature extraction for tree trunks,thereby improving segmentation accuracy.We further integrate the lightweight convolutional block attention module(CBAM),enabling the model to focus on tree trunk features while suppressing irrelevant information,which leads to additional gains in segmentation accuracy.To enable RT-Trunk to operate effectively under diverse complex forest environments,we constructed a comprehensive dataset for training and testing by combining self-collected data with multiple public datasets covering different locations,seasons,weather conditions,tree species,and levels of forest clutter.Com-pared with the other tree trunk segmentation methods,the RT-Trunk method achieved an average precision of 91.4%and the fastest inference speed of 32.9 frames per second.Overall,the proposed RT-Trunk provides superior trunk segmentation performance that balances speed and accu-racy,making it a promising solution for supporting under-canopy UAVs in the autonomous extraction of standing tree stem attributes.The code for this work is available at https://github.com/NEFU CVRG/RT Trunk.
基金funded by Thammasat University,Contract No.12/2565(TUFT 12/2565).
文摘Oil palm tree monitoring is essential to track productivity and prevent diseases,but manual methods are labor-intensive,cost-prohibitive,and error-prone.While existing remote sensing data and deep learning methods offer effi cient alternatives for monitor-ing,achieving a highly accurate tree condition classification remains a challenge.Therefore,this study classifies healthy,yellow,small,and dead oil palm trees using the YOLOv8 model.A publicly available multi-class oil palm tree dataset,after extensive correction of identified labeling errors,is used to train the model.The model's performance is compared with state-of-the-art object detectors,and a prototype web application is developed to test the model on adverse unseen image scenes.The YOLOv8 model achieves up to 99.7%F1-score and 99.3%mAP across all classes of the corrected dataset.It outperforms other object detectors and produces similar scores as the recent YOLOv10-large model.Further validation on unseen images from the developed prototype results in 76.0%F1-score and 77.9%mAP across all classes.Finally,implications empha-size the role of YoLOv8 in handling class imbalance resulting from underrepresented classes in the dataset.The experimental findings,practical demonstration,and implications presented in this paper offer robust and reliable monitoring of oil palm trees with innova-tions in precision agriculture.
文摘Purpose-Taking into consideration the current human need for agricultural produce such as rice that requires water for growth,the optimal consumption of this valuable liquid is important.Unfortunately,the traditional use of water by humans for agricultural purposes contradicts the concept of optimal consumption.Therefore,designing and implementing a mechanized irrigation system is of the highest importance.This system includes hardware equipment such as liquid altimeter sensors,valves and pumps which have a failure phenomenon as an integral part,causing faults in the system.Naturally,these faults occur at probable time intervals,and the probability function with exponential distribution is used to simulate this interval.Thus,before the implementation of such high-cost systems,its evaluation is essential during the design phase.Design/methodology/approach-The proposed approach included two main steps:offline and online.The offline phase included the simulation of the studied system(i.e.the irrigation system of paddy fields)and the acquisition of a data set for training machine learning algorithms such as decision trees to detect,locate(classification)and evaluate faults.In the online phase,C5.0 decision trees trained in the offline phase were used on a stream of data generated by the system.Findings-The proposed approach is a comprehensive online component-oriented method,which is a combination of supervisedmachine learning methods to investigate system faults.Each of thesemethods is considered a component determined by the dimensions and complexity of the case study(to discover,classify and evaluate fault tolerance).These components are placed together in the form of a process framework so that the appropriate method for each component is obtained based on comparison with other machine learning methods.As a result,depending on the conditions under study,the most efficient method is selected in the components.Before the system implementation phase,its reliability is checked by evaluating the predicted faults(in the system design phase).Therefore,this approach avoids the construction of a high-risk system.Compared to existing methods,the proposed approach is more comprehensive and has greater flexibility.Research limitations/implications-By expanding the dimensions of the problem,the model verification space grows exponentially using automata.Originality/value-Unlike the existing methods that only examine one or two aspects of fault analysis such as fault detection,classification and fault-tolerance evaluation,this paper proposes a comprehensive processoriented approach that investigates all three aspects of fault analysis concurrently.
