Project ECHIDNA (Drones)

Echidna Thermal Imagery Library

Help build a shared thermal imagery resource for echidna conservation. We are gathering raw recordings to refine human interpretation, investigate detection challenges, and support future evaluation of AI-assisted survey methods.

Contribution Criteria

What We Are Looking For

Original drone-based thermal recordings are particularly valuable, though handheld, tripod-mounted, and trail-based thermal captures are equally vital to benchmarking echidna detection.

Active & Concealed Behaviours

Paired Visible-Light Frames

False Positives & Misses

Recordings of echidnas foraging, moving, resting, curled in defense, or partially obscured beneath leaf litter, tussock grass, and hollow logs.

Dual-sensor footage, optical reference stills, or independent ground-truthing observations that confirm animal presence and posture alongside thermal tracks.

Thermal hot spots such as sun-warmed rocks and termite mounds, alongside footage of confirmed echidna locations that were missed by thermal sensors.

Australian dry sclerophyll woodland floor covered in fallen bark, hollow eucalyptus timber, and dry grasses under clear natural lighting.
Australian dry sclerophyll woodland floor covered in fallen bark, hollow eucalyptus timber, and dry grasses under clear natural lighting.

Biogeographical Scope

Taxonomic and Geographic Focus

We welcome imagery of short-beaked echidnas (Tachyglossus aculeatus) across all Australian environments and New Guinea habitats. We also welcome archival thermal observations of critically endangered long-beaked echidnas (Zaglossus spp.) in Papua New Guinea and Indonesian New Guinea.

Engagement across New Guinea represents an early, respectful research dialogue, not an established drone operations programme. All work strictly respects Indigenous custodianship, local communities, and formal research permissions. Thermal drone detection of long-beaked echidnas remains entirely unproven in the field.

• Scientific Objectives

How the Imagery Could Be Used

These activities represent exploratory conservation research objectives, not completed or guaranteed survey outcomes. Thermal sensors do not reliably penetrate dense vegetation or heavy forest canopies.

Observer Training

AI Benchmark Datasets

Detection Probability

Developing reference libraries and visual guides to train human reviewers in distinguishing subtle thermal signatures from warm background clutter.

Providing rigorous cross-sensor datasets to evaluate machine learning models and test automated detector reliability under varied field conditions.

Assessing false-positive rates, thermal dissipation curves, and probability of detection across flight altitudes, ambient heat, and vegetation cover.

Recommended Metadata Checklist

— Metadata Guidance

Useful Recording Information

Species identity and verification method: specify whether confirmed by direct visual sighting, paired optical image, or physical field sign.

When submitting footage, accompanying contextual notes allow researchers to calibrate detection models accurately against environmental variables.

Sensor and flight parameters: camera model, thermal resolution, focal length, platform type (drone, handheld, trail), flight height, and speed.

Locations may be provided at coarse bioregional resolutions or kept strictly confidential to protect sensitive populations and private properties.

Contributors retain ownership of their imagery. Permissions for sharing, research, publication and AI training will be agreed before use, and sensitive wildlife locations will be protected.

Environmental conditions: date, time of day, approximate ambient temperature, ground thermal heating, solar exposure, and canopy density.

Approximate location: bioregion, state, or coarse coordinates. Specific sensitive site details will be handled under confidentiality agreements.