Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LiDAR-Forest Dataset: LiDAR Point Cloud Simulation Dataset for Forestry Application

Yawen Lu, Zhuoyang Sun, Jinyuan Shao, Qianyu Guo, Yunhan Huang, Songlin Fei, Yingjie Chen

TL;DR

LiDAR-Forest addresses the scarcity of realistic LiDAR forest data by introducing an Unreal Engine–based simulation tool and a configurable forest dataset. The approach combines asset creation, scene generation, and LiDAR sensing in a five-module pipeline that injects sensor noise and human motion to bridge the gap to real-world data. It defines metrics such as InfraD and InfraNUC and demonstrates multiple LiDAR configurations to enable forest reconstruction, DBH estimation, and biomass assessment with error-free labels. The work enables efficient, scalable testing for forestry perception tasks and paves the way for multimodal sensor fusion and digital-twin forestry applications upon public release.

Abstract

The popularity of LiDAR devices and sensor technology has gradually empowered users from autonomous driving to forest monitoring, and research on 3D LiDAR has made remarkable progress over the years. Unlike 2D images, whose focused area is visible and rich in texture information, understanding the point distribution can help companies and researchers find better ways to develop point-based 3D applications. In this work, we contribute an unreal-based LiDAR simulation tool and a 3D simulation dataset named LiDAR-Forest, which can be used by various studies to evaluate forest reconstruction, tree DBH estimation, and point cloud compression for easy visualization. The simulation is customizable in tree species, LiDAR types and scene generation, with low cost and high efficiency.

LiDAR-Forest Dataset: LiDAR Point Cloud Simulation Dataset for Forestry Application

TL;DR

LiDAR-Forest addresses the scarcity of realistic LiDAR forest data by introducing an Unreal Engine–based simulation tool and a configurable forest dataset. The approach combines asset creation, scene generation, and LiDAR sensing in a five-module pipeline that injects sensor noise and human motion to bridge the gap to real-world data. It defines metrics such as InfraD and InfraNUC and demonstrates multiple LiDAR configurations to enable forest reconstruction, DBH estimation, and biomass assessment with error-free labels. The work enables efficient, scalable testing for forestry perception tasks and paves the way for multimodal sensor fusion and digital-twin forestry applications upon public release.

Abstract

The popularity of LiDAR devices and sensor technology has gradually empowered users from autonomous driving to forest monitoring, and research on 3D LiDAR has made remarkable progress over the years. Unlike 2D images, whose focused area is visible and rich in texture information, understanding the point distribution can help companies and researchers find better ways to develop point-based 3D applications. In this work, we contribute an unreal-based LiDAR simulation tool and a 3D simulation dataset named LiDAR-Forest, which can be used by various studies to evaluate forest reconstruction, tree DBH estimation, and point cloud compression for easy visualization. The simulation is customizable in tree species, LiDAR types and scene generation, with low cost and high efficiency.
Paper Structure (9 sections, 2 equations, 4 figures)

This paper contains 9 sections, 2 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Background
  3. LiDAR-Forest Dataset
  4. Dataset Simulation
  5. Metrics and Evaluations
  6. Potential Applications
  7. Future works
  8. Conclusion
  9. Acknowledgments

Figures (4)

  • Figure 1: Overview of our newly proposed LiDAR-Forest simulation dataset. The simulation system consists of asset creation, scene generation, and LiDAR point cloud generation. The system is able to benefit in a variety of forestry scenarios, such as tree species identification, stem mapping and measuring, and forest digital twinning. Note: The generated point cloud specifies different colors for tree leaves (red), tree stems (green), and landscape (yellow). The project page will be available at: https://lidar-simulate.github.io/LiDAR_simulate/.
  • Figure 2: Conceptual difference between the real-word LiDAR backpack scanner (left) and our simulated LiDAR scanner in wild forests (right).
  • Figure 3: Specific embodiment of the two error simulation methods: Option 1 corresponds to the scheme that modifies the angle, and Option 2 corresponds to the scheme modifying the coordinates directly.
  • Figure 4: Illustration of the simulation results from different types of LiDAR sensors in our LiDAR-Forest dataset. Top left: simulation result from 8-bit LiDAR; Top right: simulation result from 16-bit LiDAR; Bottom left: simulation from 64-bit LiDAR; Bottom right: simulation from 256-bit LiDAR. All the given point clouds are generated from the same scenario and the same viewpoint using different sensor patterns.