Detecting crop pests with artificial intelligence

Silverleaf whitefly detection apps under trial

Silverleaf whitefly reduces the end value of cotton due to their sticky honeydew excretions contaminating the lint and preventing it from being processed.

Silverleaf whitefly reduces the end value of cotton due to their sticky honeydew excretions contaminating the lint and preventing it from being processed.


USQ are developing a mobile app to stop pests from flying under the radar.


Artificial intelligence is being employed to help Australian cotton farmers stop a major pest from flying under the radar.

Near impossible to identify and count with the naked eye, the silverleaf whitefly has increased in prevalence in recent years, considerably reducing the end value of the crop thanks to their sticky honeydew excretions contaminating cotton lint and preventing it from being processed.

Led by University of Southern Queensland researchers Dr Alison McCarthy and Dr Derek Long, in will collaboration with Queensland Department of Agriculture and Fisheries researcher Dr Paul Grundy, the project focus is developing a new artificial intelligence smartphone app.

The project has received funding from the Cotton Research and Development Corporation.

"Traditionally, sampling is labour-intensive and done manually, with growers and their agronomists having to closely monitor the changes in the numbers of pests across hundreds of cotton plant leaves on a weekly basis to determine if control action is required," Dr McCarthy said.

"We identified that machine vision could automate the pest counting on each leaf by using infield cameras and image analysis software. We have since enabled these vision detection algorithms to be used on a smartphone device.

"Through an app, agronomists can then use real-time photo capture for pest counting which offers reduced sampling times, more precise detection and recording of pests, increased sampling consistency between field personnel and improvement for the timing of control decisions."

The first version of the app was tested by agronomists and researchers in the 2019/20 season in two cotton growing regions.

The project's principal researcher Dr Derek Long said feedback from first time users of the app was helping the research team design the next version.

"Agronomists responded positively to the logging capability in the app and with further refinements being incorporated into the second version. We expect the app to be much faster than manually counting whiteflies and referencing threshold advice contained in the industry's Pest Management Guide," he said.

St George district crop consultant Jamie Street participated in the 2019/20 trial and said once the app was commercially viable, it had great potential going forward.

"I found it very user friendly and it is a much quicker way to access the silverleaf whitefly than doing it by the naked eye," Mr Street said.

Dr Long said refining the app to detect a range of cotton insects would further aid agronomists and better inform pest management decisions.

An updated version will be released for testing during the 2020/21 cotton season.


From the front page

Sponsored by