Detecting Insects is Getting High-Tech on UBC Campus
A new Campus as a Living Laboratory project will enable UBC researchers to monitor insects in real-time, on a scale never done before.
“This project promises to help us understand insect biodiversity, which we know is both crucial to ecosystems, and affected by human activities,” says Juli Carrillo, Assistant Professor in the Faculty of Land and Food Systems, and lead researcher of the Digital Detection Web for On-Campus Insects project.
Understanding insects in real-time
Monitoring insects has been a time-consuming and inaccurate process. Farmers and scientists capture a small fraction using sticky cards – simple sheets of paper with glue that are placed out in the field. Traditionally, these are retrieved weekly and the trapped insects are manually identified. In order to dramatically improve the process, Postdoctoral Fellow Quentin Geissmann has developed a technology called “Sticky Pi” that automates this monitoring and scales it up significantly.
Sticky Pi is a “smart” trap with a camera that takes pictures every 20 minutes. It uses deep learning – a subfield of machine learning using algorithms – to automatically identify insects, in real-time.
“There are two main impacts of scaling up insect biodiversity monitoring.” says Geissmann. “First, we save many precious hours of work through automation. Second, and more importantly, because we acquire data in real-time, we can see new patterns. For instance, we are very interested in the time of the day insects are active and how they respond to weather variations.”
More than 50 Sticky Pi traps will be deployed across UBC campus, from residential areas to the fields and forest of the UBC Farm, allowing for a new understanding of insect biodiversity in these different ecosystems.
A pilot program using Sticky Pi has already had an impact beyond UBC, showing its potential to serve as an early alert system for new pests. The team confirmed the arrival of an invasive insect that is native to Europe, the strawberry blossom weevil, which damages berry crops by feeding on the flower buds of host plants. The weevil is listed as an urgent concern for Canada and can affect trade policy with the U.S.
Researchers across the globe are invited to use the technology and deep learning tools, which are open-source and open-hardware: “We built the hardware on top of the fantastic Raspberry Pi microcomputer and used 3D-printing, which keeps our work accessible,” says Geissmann. “People all over the world can freely build, use and adapt the different parts.”
Involving the UBC campus community
Now the researchers are inviting the UBC campus community to join the project.
Community members are encouraged to “adopt a Sticky Pi” by using a team-built mobile app to ‘harvest’ the data themselves. The team will host entomology workshops where community members will learn and help annotate images of insects, which trains the algorithm. The team will also connect diverse groups from farming and computer sciences through “hackathons” to prototype original solutions to problems in biodiversity and agriculture.
This project will not only create baseline data for UBC Campus and the UBC Farm on insect biodiversity but will monitor change and improve future management decisions.
There are implications for many different communities.
“This project addresses the pest monitoring needs of growers and improves existing technology for researchers,” says Carrillo. “But it also benefits the greater community by increasing our understanding of the land and space in which we live.”
Find out more about the project at https://sustain.ubc.ca/digital-detection-web-campus-insects