Automated thermal imagery analysis platform for multiple pest species

AI feral deer pigs software

Status: Completed

Start date: 1 July 2020

Completion date: 30 September 2022

Project code: P01-T-003

Species/Threats: Feral deer, Feral pigs

Download project report (PDF, 3.83 MB)

Summary

Designed for ecologists and land managers, ThermEye is a tool that uses thermal imagery and artificial intelligence to differentiate between different vertebrate pests during monitoring or aerial culling.
ThermEye can identify and differentiate between sheep, cattle, deer, goats, rabbits and kangaroos. Its algorithm was trained, then proven in the field, and re-trained with more data.

Objectives

The objective of this project was:

  1. Develop a deep machine learning model for fully autonmous thermal imagery analysis for multiple pest animal species and demonstrate the application for managing invasvie species on a landscape level.

Key achievements

Outputs

  • Automated thermal analysis algorithm.
  • Thermal vision libraries for pest species.
  • Software package, ‘ThermEye’, automated analysis platform.

Outcomes

  • Greater ability for on ground managers to monitor and measure density of pest populations.
  • Greater ability to measure the effect of control operations.

Impact

  • Increased effectiveness of monitoring and surveillance for invasive species over large land areas.
  • Reduced impact costs of invasive species through more effective management and improved monitoring.
  • More effective resource allocation for incursion responses due to increased
    capability in large area monitoring and surveillance.
  • Reduced total impact costs of invasive species incursions by improving detection that allows for more effective responses.
  • Contribution to reduced negative environmental impacts from invasive species through more effective management helped by improved monitoring and surveillance.
  • Increased scientific knowledge and research capacity developed for thermal imaging and automated analysis algorithms.

Project team

Dr Peter Adams

Project Lead | WA DPIRD

Dr Stuart Dawson

WA DPIRD

Tom Low

Tomcat Technologies

Anwaar Ul-Haq

Charles Sturt University

Asim Khan

Victoria University

Paul Manoranjan

Charles Sturt University

Dr Tarnya Cox

NSW DPI

Dr Matt Gentle

QDAF

Project partners

Project partners including Tomcat Technologies. The project received funding from the Australian Government Department of Agriculture, Fisheries and Forestry (DAFF).

Project updates

February 2021

A proof-of-concept AI software model for detecting multiple species has been achieved. This model has been trained to detect and identify pigs, rabbits and deer from thermal imagery. Further exposure to thermal imagery of these species is integral to training the AI. End user consultations have revealed a need for this software to be deployed in real time during pest animal surveys.

Scientific publications & reports

Ulhaq A, Adams P, Cox TE, Khan A, Low T and Paul M (2021) Automated Detection of Animals in Low-Resolution Airborne Thermal Imagery Remote Sensing 13(16). https://doi.org/10.3390/rs13163276

More like this

New

WeedRemeed™: Drone AI detection of priority weed species at scale
Completed

National registration with the APVMA of a feral cat bait Stage 1
In Progress

Facilitating community adoption of digital resources (FeralScan)
VIEW ALL RESEARCH PROJECTS