{"title":"Advancing tree detection in forest environments: A deep learning object detector approach with UAV LiDAR data","authors":"Sina Jarahizadeh, Bahram Salehi","doi":"10.1016/j.ufug.2025.128695","DOIUrl":null,"url":null,"abstract":"<div><div>The initial phase in determining tree parameters within urban and forest environments is Individual Tree Detection (ITD), which includes tree count, spatial distribution, height, volume, crown dimensions, and species identification. This process holds significance in applications like urban forest inventory, planning, and tree carbon accounting. Traditional airborne and spaceborne remote sensing data lack the precision required for ITD due to their coarse spatial resolution. High-resolution multispectral and Light Detection and Ranging (LiDAR) data collected by Unmanned Aerial Vehicle (UAV) sensors have enabled detailed ITD and tree parameter estimation. However, the processing of such very high-resolution data presents challenges. While existing algorithms for processing 2-dimensional (2D) and 3-dimensional (3D) data from airborne sensors exist, they prove impractical for UAV data, primarily due to its extremely high spatial resolution. Recent strides in deep-learning algorithms offer promising solutions for ITD using UAV data. This paper introduces a novel ITD method using a modified You Only Look Once V7 (YOLO V7) deep learning object detection framework, employing UAV LiDAR data. The approach involves rasterizing point clouds in various channels, including Vertical Density (VD), Canopy Height Model (CHM), Gradient of the CHM (G-CHM), and Local Binary Pattern of the CHM (LBP-CHM). Subsequently, the YOLO V7 object detector is employed to identify the bounding box of each tree. The modified YOLO7 algorithm is trained and tested on UAV LiDAR data collected over diverse regions of interest, encompassing pine, deciduous, and mixed tree types with varying tree densities. The results exhibit a substantial enhancement over the previously developed YOLO3 on airborne LiDAR data, showcasing heightened accuracy, precision, and recall within the ranges of 0.7–0.94, 0.76–0.99, and 0.8–0.97, respectively. From a practical standpoint, our automated method holds potential for urban tree inventory updates and serves as a valuable tool for ground-truthing large-scale satellite-based forest structure and biomass estimation, among various other applications.</div></div>","PeriodicalId":49394,"journal":{"name":"Urban Forestry & Urban Greening","volume":"105 ","pages":"Article 128695"},"PeriodicalIF":6.0000,"publicationDate":"2025-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Urban Forestry & Urban Greening","FirstCategoryId":"93","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1618866725000299","RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENVIRONMENTAL STUDIES","Score":null,"Total":0}
引用次数: 0
Abstract
The initial phase in determining tree parameters within urban and forest environments is Individual Tree Detection (ITD), which includes tree count, spatial distribution, height, volume, crown dimensions, and species identification. This process holds significance in applications like urban forest inventory, planning, and tree carbon accounting. Traditional airborne and spaceborne remote sensing data lack the precision required for ITD due to their coarse spatial resolution. High-resolution multispectral and Light Detection and Ranging (LiDAR) data collected by Unmanned Aerial Vehicle (UAV) sensors have enabled detailed ITD and tree parameter estimation. However, the processing of such very high-resolution data presents challenges. While existing algorithms for processing 2-dimensional (2D) and 3-dimensional (3D) data from airborne sensors exist, they prove impractical for UAV data, primarily due to its extremely high spatial resolution. Recent strides in deep-learning algorithms offer promising solutions for ITD using UAV data. This paper introduces a novel ITD method using a modified You Only Look Once V7 (YOLO V7) deep learning object detection framework, employing UAV LiDAR data. The approach involves rasterizing point clouds in various channels, including Vertical Density (VD), Canopy Height Model (CHM), Gradient of the CHM (G-CHM), and Local Binary Pattern of the CHM (LBP-CHM). Subsequently, the YOLO V7 object detector is employed to identify the bounding box of each tree. The modified YOLO7 algorithm is trained and tested on UAV LiDAR data collected over diverse regions of interest, encompassing pine, deciduous, and mixed tree types with varying tree densities. The results exhibit a substantial enhancement over the previously developed YOLO3 on airborne LiDAR data, showcasing heightened accuracy, precision, and recall within the ranges of 0.7–0.94, 0.76–0.99, and 0.8–0.97, respectively. From a practical standpoint, our automated method holds potential for urban tree inventory updates and serves as a valuable tool for ground-truthing large-scale satellite-based forest structure and biomass estimation, among various other applications.
期刊介绍:
Urban Forestry and Urban Greening is a refereed, international journal aimed at presenting high-quality research with urban and peri-urban woody and non-woody vegetation and its use, planning, design, establishment and management as its main topics. Urban Forestry and Urban Greening concentrates on all tree-dominated (as joint together in the urban forest) as well as other green resources in and around urban areas, such as woodlands, public and private urban parks and gardens, urban nature areas, street tree and square plantations, botanical gardens and cemeteries.
The journal welcomes basic and applied research papers, as well as review papers and short communications. Contributions should focus on one or more of the following aspects:
-Form and functions of urban forests and other vegetation, including aspects of urban ecology.
-Policy-making, planning and design related to urban forests and other vegetation.
-Selection and establishment of tree resources and other vegetation for urban environments.
-Management of urban forests and other vegetation.
Original contributions of a high academic standard are invited from a wide range of disciplines and fields, including forestry, biology, horticulture, arboriculture, landscape ecology, pathology, soil science, hydrology, landscape architecture, landscape planning, urban planning and design, economics, sociology, environmental psychology, public health, and education.