Article
Environmental Sciences
Zongqi Yao, Guoqi Chai, Lingting Lei, Xiang Jia, Xiaoli Zhang
Summary: The use of UAV hyperspectral images and the Mask R-CNN model enables automatic, efficient, and accurate identification of individual tree species and extraction of crown parameters, providing practical technical support for forest management and ecological diversity monitoring.
Article
Environmental Sciences
Chong Zhang, Jiawei Zhou, Huiwen Wang, Tianyi Tan, Mengchen Cui, Zilu Huang, Pei Wang, Li Zhang
Summary: High-resolution UAV imagery combined with a convolutional neural network approach is effective in accurately measuring forestry ecosystems. In this study, a new method for individual tree segmentation and identification based on the improved Mask R-CNN is proposed, which shows advantages in broadleaf canopy segmentation and number detection.
Article
Ecology
James G. C. Ball, Sebastian H. M. Hickman, Tobias D. D. Jackson, Xian Jing Koay, James Hirst, William Jay, Matthew Archer, Melaine Aubry-Kientz, Gregoire Vincent, David A. A. Coomes
Summary: Tropical forests play a crucial role in the global carbon cycle and biodiversity, but estimating the number of large trees is challenging. In this study, a deep learning method called Detectree2 was developed to automatically recognize and segment large trees using aerial imagery. This method has great potential in various applications such as forest ecology and conservation.
REMOTE SENSING IN ECOLOGY AND CONSERVATION
(2023)
Article
Geography, Physical
Zhenbang Hao, Lili Lin, Christopher J. Post, Elena A. Mikhailova, Minghui Li, Yan Chen, Kunyong Yu, Jian Liu
Summary: This study developed a Mask R-CNN model to automatically detect tree crown and height of Chinese fir in a plantation, achieving high accuracy. Results showed that CHM input images outperformed DSM images for tree crown delineation and height assessment, highlighting the potential of Mask R-CNN in assisting remote sensing applications in forests.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2021)
Article
Forestry
Maojia Gong, Weili Kou, Ning Lu, Yue Chen, Yongke Sun, Hongyan Lai, Bangqian Chen, Juan Wang, Chao Li
Summary: This study used high-resolution UAV imagery and the Mask R-CNN deep learning model to analyze the crown width of Malania oleifera trees, established a growth model between crown width and diameter at breast height, and predicted the aboveground biomass of individual M. oleifera based on the empirical equation. This study provides a reference method for estimating the aboveground biomass of individual M. oleifera trees.
Article
Environmental Sciences
Yingbo Li, Guoqi Chai, Yueting Wang, Lingting Lei, Xiaoli Zhang
Summary: Accurate and automatic identification of tree species at the individual tree scale is important but rarely explored in complex forest environments. In this study, the authors propose ACNet and ACE R-CNN for tree species identification, which selectively fuse features from RGB images and CHM data, and improve edge accuracy through edge loss and Sobel filter. Experimental results demonstrate the effectiveness of the proposed method.
Article
Environmental Sciences
Rao Fu, Jing He, Gang Liu, Weile Li, Jiaqi Mao, Minhui He, Yuanyang Lin
Summary: The purpose of this study is to quickly determine the extent and size of post-earthquake seismic landslides using a small amount of post-earthquake seismic landslide imagery data. Different backbone networks were used for training and identification, and the performance of the improved model was significantly better in terms of accuracy and recognition in Haiti's post-earthquake images.
Article
Chemistry, Multidisciplinary
Peichao Cong, Jiachao Zhou, Shanda Li, Kunfeng Lv, Hao Feng
Summary: This study proposes a feature-map-based MSEU R-CNN method for citrus tree crown segmentation, which achieves accurate segmentation by using pixel-aligned and visually distance-adjusted RGB-D images. The method utilizes squeeze-and-excitation (SE) and UNet++ to fully fuse feature map information and achieves near-real time detection speed. It outperforms previous-best Mask R-CNN in terms of segmentation accuracy and speed.
APPLIED SCIENCES-BASEL
(2023)
Article
Environmental Sciences
Xingmei Xu, Lu Wang, Meiyan Shu, Xuewen Liang, Abu Zar Ghafoor, Yunling Liu, Yuntao Ma, Jinyu Zhu
Summary: This study proposed a method for detecting and counting maize leaves based on deep learning with RGB images collected by unmanned aerial vehicles (UAVs). The Mask R-CNN was used for segmentation and YOLOv5 for leaf detection. The results showed improved segmentation performance and leaf counting accuracy. The use of UAV images for field-grown crop leaf counting research was found to be promising.
Article
Environmental Sciences
Yi Gan, Quan Wang, Atsuhiro Iio
Summary: The automatic detection of tree crowns and estimation of crown areas using RGB imagery from unmanned aerial vehicles (UAVs) have significant implications for biodiversity and ecosystem conservation. Two deep-learning-based models, Detectree2 and DeepForest, were evaluated for their effectiveness in detecting tree crowns in an alpine deciduous forest. Detectree2 outperformed DeepForest and both models showed potential for successful tree crown prediction, with the spatial resolution having an impact on detection accuracy.
Article
Environmental Sciences
Kunyong Yu, Zhenbang Hao, Christopher J. Post, Elena A. Mikhailova, Lili Lin, Gejin Zhao, Shangfeng Tian, Jian Liu
Summary: This study evaluated the accuracy of different algorithms for detecting individual trees using remote sensing images. The results showed that the Mask R-CNN algorithm achieved the highest accuracy, followed by the LM algorithm and MCWS algorithm. The study also highlighted the requirements and limitations of each algorithm, with Mask R-CNN requiring additional training data and computational resources. Providing valuable information for selecting optimal approaches for detecting individual trees, this study contributes to the field of precision forest management.
Article
Ecology
Jeremy Arkin, Nicholas C. C. Coops, Lori D. D. Daniels, Andrew Plowright
Summary: This article introduces a new method that uses RPAS-acquired data to accurately characterize the impact of wildfires on individual trees. By calculating the crown scorch height and volume, the severity of fire damage to trees can be assessed. The research results show that this method can classify and extract the attributes of burnt trees with high accuracy.
Article
Environmental Sciences
Mirela Beloiu, Lucca Heinzmann, Nataliia Rehush, Arthur Gessler, Verena C. Griess
Summary: The study successfully used the Convolutional Neural Network algorithm, Faster R-CNN, and open-source aerial RGB imagery to geolocate and identify four tree species in heterogeneous forests. The average detection accuracy of single-species models was 0.76, and the accuracy increased in multi-species models. The performance of the models was mainly influenced by forest stand structure.
Article
Forestry
Zhenyu Wu, Xiangtao Jiang
Summary: A Mask R-CNN-based algorithm is proposed in this paper for pine wilt disease detection and extraction, which can accurately detect the disease and extract the infected regions. Experimental results show that the proposed method can effectively identify the distribution of diseased pine trees and calculate the damage proportion in a relatively accurate way, facilitating forest management.
Article
Engineering, Civil
Junyeon Chung, Hoon Sohn
Summary: This study quantitatively measures the exposed shank length of a bolt using an RGB-depth camera and a mask-region-based convolutional neural network, without requiring any measurement from the initial state of the bolt. The proposed technique successfully detects bolt loosening over 3 mm with a resolution of 1 mm and 97% accuracy at different camera angles and distances.
SMART STRUCTURES AND SYSTEMS
(2021)
Article
Computer Science, Information Systems
Davit Marikyan, Savvas Papagiannidis, Eleftherios Alamanos
Summary: This study addresses the outcomes of technology use when it falls short of expectations and the coping mechanisms users may use in such circumstances. By adopting Cognitive Dissonance Theory, the study explores how negative disconfirmation of expectations can result in positive outcomes and how negative emotions impact the selection of dissonance reduction mechanisms. The study finds that post-disconfirmation dissonance leads to feelings of anger, guilt, and regret, which correlate with dissonance reduction mechanisms, ultimately affecting satisfaction and well-being.
INFORMATION SYSTEMS FRONTIERS
(2023)
Article
Biodiversity Conservation
Paul W. Hacker, Nicholas C. Coops, Etienne Laliberte, Sean T. Michaletz
Summary: The association between leaf chemicals and reflectance values can be used to model and predict individual functional traits. The accuracy of prediction is affected by spectral mixing, particularly for traits such as percent nitrogen and equivalent water thickness. Species-specific relationships are more important for nitrogen content and water thickness, while a general model can be more applicable for traits like chlorophyll concentration and leaf mass per area. Different combinations of plant species in a mixed spectrum result in varied prediction errors.
ECOLOGICAL INDICATORS
(2022)
Article
Forestry
Nicholas C. Coops, Piotr Tompalski, Tristan R. H. Goodbody, Alexis Achim, Christopher Mulverhill
Summary: This article aims to develop a conceptual framework for forestry inventory update, known as a 'living inventory'. The framework includes the critical components of inventory and growth monitoring, change detection, and error propagation. By integrating advanced remote sensing data and satellite data, it provides methods for updating forest condition information, predicting future growth and yield, and guiding forest management and silvicultural decisions.
Article
Environmental Sciences
Spencer Dakin Kuiper, Nicholas C. Coops, Piotr Tompalski, Scott G. Hinch, Alyssa Nonis, Joanne C. White, Jeffery Hamilton, Donald J. Davis
Summary: Understanding changes in salmonid populations and their habitat is crucial due to climate change and their importance as a keystone species. Airborne Laser Scanning (ALS) data can be used to assess the quality and quantity of salmonid habitat, as well as characterize detailed stream attributes. ALS data provides detailed Digital Elevation Models (DEMs) and can be utilized for sustainable forest management decision making and advanced salmonid habitat modeling.
REMOTE SENSING OF ENVIRONMENT
(2022)
Article
Environmental Sciences
Christopher Mulverhill, Nicholas C. Coops, Txomin Hermosilla, Joanne C. White, Michael A. Wulder
Summary: The study evaluated the agreement between two broad-scale forest canopy height products across different ecological gradients. Overall, the two datasets showed high correspondence, but there were variations in agreement in different ecozones. The study also found that the modeled heights based on optical satellite data had a less generalized distribution than heights from ICESat-2.
REMOTE SENSING OF ENVIRONMENT
(2022)
Article
Forestry
Rohit Arora, Taraneh Sowlati, Joel Mortyn, Dominik Roeser, Verena C. Griess
Summary: A mixed-integer linear programming model is developed in this paper to optimize the scheduling of harvesting activities, considering the precedence relationship among activities. The objective is to minimize the total costs by determining the start and end time of each activity at each cut block while considering machine movement time. The model is applied to a case study of a large forest company in British Columbia, Canada, with the result of a harvesting cost only 1.37% higher than the lowest possible cost and only 3 idle machines. A detailed harvesting schedule is generated based on the start and end time and operating time of each activity at each cut block.
INTERNATIONAL JOURNAL OF FOREST ENGINEERING
(2023)
Article
Remote Sensing
Martin Queinnec, Nicholas C. Coops, Joanne C. White, Verena C. Griess, Naomi B. Schwartz, Grant McCartney
Summary: In this study, dominant species groups in a large boreal forest were mapped by combining area-based and individual tree metrics derived from LiDAR data with multispectral information from Sentinel-2 imagery. The study found that variables such as reflectance in the red edge region, tree crown area and volume, and cumulative distribution of LiDAR returns in the canopy were important for discriminating between species groups.
CANADIAN JOURNAL OF REMOTE SENSING
(2023)
Article
Remote Sensing
Levi Keay, Christopher Mulverhill, Nicholas C. C. Coops, Grant McCartney
Summary: The advent of CubeSat constellations has revolutionized the ability to observe Earth systems through time. This study developed and implemented a method for the spatial and temporal detection of forest harvest operations using images from the PlanetScope constellation. Results indicate that forest harvesting can be detected with relative accuracy, providing previously unavailable levels of spatial and temporal detail for forest stakeholders.
CANADIAN JOURNAL OF REMOTE SENSING
(2023)
Article
Remote Sensing
Alexandre Morin-Bernard, Alexis Achim, Nicholas C. Coops
Summary: Non-stand-replacing disturbances play a significant role in northern hardwood forest dynamics, but are more difficult to characterize using satellite imagery than stand-replacing events. This study proposes a hurdle approach that attributes disturbance causal agents to specific sample plots, achieving an overall accuracy of 82.9%. Disturbance-specific models were then developed to assess the severity of partial harvests and damage from ice storms, with r-squared values of 0.57 and 0.59, respectively. These models provide important information for future silvicultural planning by capturing within-stand variability in disturbance severity.
CANADIAN JOURNAL OF REMOTE SENSING
(2023)
Review
Fisheries
Spencer Dakin Kuiper, Nicholas C. C. Coops, Scott G. G. Hinch, Joanne C. C. White
Summary: Remote sensing technology has the potential to revolutionize freshwater fish habitat monitoring by providing information across large geographic areas, but the overwhelming number of platforms, sensors, and software available may hinder its widespread use. This review examines the fundamental characteristics of remote sensing technologies used for freshwater habitat characterization, reviews studies that have utilized these technologies, and identifies key habitat features, fish species, and regions that have been examined. The review also highlights the strengths and weaknesses of different remote sensing technologies, suggests future research directions, and provides important considerations for those interested in utilizing these technologies for freshwater fish habitat characterization.
FISH AND FISHERIES
(2023)
Article
Environmental Sciences
Camilla Moioli, Anil Shrestha, Dominik Roeser, Guangyu Wang, Terry Sunderland, Hisham Zerriffi
Summary: Despite global momentum in restoration activities, little attention has been paid to the socio-economic implications, with limited evidence on equity and equality outcomes. This study focuses on investigating the fairness within the Chinese Conversion of Cropland to Forest Program (CCFP) and proposes a quantitative methodology to assess equity and equality. The findings reveal a shift in households' economic structure, with a decrease in farming activities and an increase in out-migration, particularly among the lowest income groups. Both equality and equity have improved, with the best outcomes in regions where the CCFP has been implemented for a longer time. The level of participation in the Program also plays a significant role in explaining income variations.
LAND DEGRADATION & DEVELOPMENT
(2023)
Article
Forestry
A. R. Wotherspoon, A. Achim, N. C. Coops
Summary: This study examines the future climate trends in eight ecozones in Canada that contain managed forests. The projections suggest a warming trend and an overall increase in precipitation. The study highlights the potential impacts on dominant species and wood volume for Canada's forestry industry.
CANADIAN JOURNAL OF FOREST RESEARCH
(2023)
Article
Forestry
Jose Riofrio, Joanne C. White, Piotr Tompalski, Nicholas C. Coops, Michael A. Wulder
Summary: By developing age-independent height growth models, using multi-temporal airborne laser scanning (ALS) data, a comprehensive indicator of site quality for complex and irregular stand structures is provided. This approach leverages the accurate, spatially detailed characterization of canopy heights afforded by ALS data and is independent of stand age, increasing the possible geographic extent of height growth estimates.
FOREST ECOLOGY AND MANAGEMENT
(2023)
Article
Geography, Physical
Saverio Francini, Txomin Hermosilla, Nicholas C. Coops, Michael A. Wulder, Joanne C. White, Gherardo Chirici
Summary: Remote sensing is a major source of information for monitoring forest dynamics, but accurate surface reflectance data is often difficult to obtain. Pixel-based composites are used to generate complete coverage of the area of interest from multi-temporal images, but a comprehensive methodology for assessing the quality of these composites is currently lacking. In this study, a pixel-based composite assessment methodology based on five criteria was introduced and tested on Landsat images over Europe. The results showed that the assessment approach was effective for evaluating the quality of pixel-based composites and could be applied in various applications.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Ecology
Margaret E. Andrew, Douglas K. Bolton, Gregory J. M. Rickbeil, Nicholas C. Coops
Summary: This study evaluates the effects of niche-based mechanisms, including environmental filtering, niche availability, and niche packing, on biodiversity patterns. The results show that the importance of these mechanisms varies with scale, position on environmental gradients, and taxonomic group.
JOURNAL OF BIOGEOGRAPHY
(2023)