Plot grid generation using machine learning with multispectral images
A machine learning solution for generating georeferenced plot grids from multispectral images, integrating into UAV workflows for enhanced agricultural monitoring and management.
Overview
The proposed solution leverages machine learning to automatically generate plot grids from multispectral orthomosaics. This technology aims to enhance UAV image processing workflows by providing georeferenced vector grids that can be overlaid on orthomosaics. By utilizing spectral features and prior knowledge of plot shapes and dimensions, this solution offers a robust approach to plot delineation and classification, which is crucial for agricultural monitoring and land management.
Technical specifications
- Input: Multispectral orthomosaics
- Output: Georeferenced vector grids
- Features: Combines spectral feature learning with prior plot knowledge for accurate grid generation
- Model: Machine learning model capable of supervised or self-supervised training based on available datasets
- Integration: Seamlessly integrates into existing UAV image processing workflows
- Validation: Relies on manually created plot grids for model training and validation
Technology readiness level
Currently at TRL 3, the solution includes a software prototype as a proof of concept. The project is in the development phase, with tasks focused on model creation, training, integration, and reporting scheduled over a six-month period.
About The University of Melbourne
The University of Melbourne is a comprehensive institution spanning STEM, health and clinical practice, business and law, and the creative and social disciplines. Co‑located hospital and research precincts, together with an inner‑city innovation ecosystem, place companies, startups, and researchers in shared labs, prototyping spaces, and studios. Engineering and technology programs connect with advanced manufacturing facilities, while structured industry projects and placements link partners with talent and translational problem‑solving. Work is supported by competitive funding from the Australian Research Council and the National Health and Medical Research Council, with additional backing from state programs and industry partners; a dedicated technology transfer office manages IP, licensing, and startup formation.
delivered to your inbox.