Development of a Ground-Based AI Vision System for Fine-Scale Weed Detection in Tropical Reforestation
Role
Principal Investigator
Description
Tropical reforestation initiatives face significant early-stage threats from fast-growing weed species that compete with young seedlings for sunlight, nutrients, and water. While drone-based monitoring provides valuable aerial perspectives, it often lacks the resolution needed to detect weeds growing in close proximity to seedlings or partially hidden by surrounding vegetation. To complement aerial detection, this project proposes the development of a ground-based computer vision system for fine-scale weed detection and growth dominance estimation, enabling precision weeding and decision support in reforestation programs. This first year of the TerraWeed project will focus on hardware setup, initial dataset creation, and the development of deep learning models for close-range classification and segmentation of weeds versus tree seedlings. This foundational work will provide a basis for deploying intelligent ground-based systems that support high-precision intervention in complex tropical environments.
Date
09 Jan 2026 - 08 Jan 2027
Project Type
N/A
Keywords
Object Detection;Deep Learning;Weed Detection;Reforestation
Funding Body
Queensland Government Department of Environment, Tourism, Science and Innovation (QDETSI), Native Conifer Carbon Sink Pty Ltd
Amount
92193.8
Project Team
Euijoon Ahn;Eric Wang;Tao Huang;Mostafa Rahimi Azghadi