Managing cotton pests such as silver leaf whitefly and cotton aphids is a time-consuming process involving manually counting pests on leaves sampled from cotton fields. These pests can cause yield loss through plant feeding as well as lint contamination from waste secretions making it essential to monitor fields carefully to determine if control actions are required.
To reduce sampling times for growers and allow more precise detection, the University of Southern Queensland (USQ) and Department of Agriculture and Fisheries (DAF) are developing a silverleaf whitefly and cotton aphid detection app for use on smartphones. This collaborative project is a unique pairing of DAF entomologists with USQ mechatronic engineers to exploit recent advances with machine learning and smart phone technology to solve a challenging biological based problem.
DAF scientist Paul Grundy said the app would enhance pest management decision-making for cotton growers and agronomists putting pest detection directly in the palm of their hands.
The app utilises the camera in your smart phone to take images of the undersides of cotton leaves upon which a machine-learning algorithm automatically detects and counts the number of viable whitefly and aphids that might be present. A prototype of this app for whitefly was deployed during the 2019-20 cotton season to a small test group of 8-10 agronomists. Their user experiences are now being utilised to improve both the underlying algorithms and app design.
‘The second version of the app, which will be released as a closed Beta version for further testing in the 2020-21 season, will provide a more interactive experience logging pest numbers over time and alerting the user when pest numbers within the crop are approaching actionable control thresholds,’ Paul said.
The new app will incorporate updated cotton aphid detection algorithms and sub-classify sliver whitefly nymphs into key groups (i.e. healthy, dead, emerged). This additional information will further aid agronomists and better inform pest control management decisions.
Paul said there is potential for this technology to be transferred across other crops, such as various vegetables, that are afflicted by whitefly and aphid pests.
Benefits for growers
Manual detection of pests is a time-consuming and menial process requiring 20-30 leaves to be sampled per 25 hectares of cotton. Crop agronomists sample and examine hundreds of leaves each week, which must be examined by eye for the presence and density of each pest.
The smartphone app aims to reduce sampling times, whilst standardising the precision of pest detection and record keeping. Ensuring consistency is particularly useful for agronomists that have teams of people who conduct seasonal checking. Potential also exists to use the data collected and generated to inform pest management decisions across larger scales in both space and time. Pest management is often at its most successful when centred around area wide strategies that account for fluctuations across farms and regions.
‘The embedded processing on the smartphone app will allow agronomists to reduce the time spent pausing to examine and manually record insect presence on each leaf allowing them to focus on crop management rather than insect counting,’ said Paul.
‘Using the app would enable larger samples to be taken in the same time period, or the same number of samples more efficiently, both of which would better underpin greater accuracy for decision-making.’
In 2020, initial results were shared with industry and the scientific community. An expression of interest is currently being developed to attract a commercial partner for the app.
Contact: Paul Grundy via email at Paul.Grundy@daf.qld.gov.au or Derek Long at Derek.Long@usq.edu.au
Partners: University of Southern Queensland. Funded by Cotton and Research Development Corporation.
Tech type: Intelligent Apps, Artificial Intelligence, Sensors