Our third and final Innovation Ignition Grant project takes on a very practical bottleneck in forest biosecurity surveillance: there’s often more trap material than there are expert hours to examine it.
The challenge: Programs like Forest Watch Australia rely on panel traps to catch insects for biosecurity surveillance — but a single program can generate over 1,270 individual trap catches, each one potentially containing hundreds of insects, plant fragments, and debris, all needing careful expert examination under a microscope before a diagnosis can be confirmed. That manual sorting process is slow, and delays in prioritising samples mean slower responses if an exotic pest turns up.
The approach: Shiron Thalagala, Yufan Zheng, and Sylvia Jepkemboi are building “Bio-SORT” — an optical AI screening system that photographs unsorted trap contents (no need to manually separate insects from debris first) and uses machine learning to flag the likely presence of priority insect genera. Each sample gets a risk score and is triaged into one of three tiers: high-risk samples go straight to expert lab diagnosis, low-risk samples move to routine processing, and anything the model is uncertain about gets flagged for expert review rather than risking a missed detection. Importantly, the system is designed to support expert diagnosis, not replace it — its whole job is to help specialists spend their time where it matters most.
They’re partnering with state governments in SA and NSW, who are providing real trap samples and lures and will help test the prototype against genuine operational conditions, ranging from simple to heavily cluttered samples.
Why it matters: If the approach proves reliable, it could meaningfully cut the time experts spend screening non-target material, freeing them up to focus on the samples that actually need a closer look — a real efficiency gain for exotic pest surveillance. Success here would also build the evidence base needed to pursue a larger operational trial and, potentially, other biosecurity agencies.
Funding awarded: $19,800
Congratulations to Shiron, Yufan and Sylvia — we can’t wait to see Bio-SORT in action.