Protecting fruit orchards from dangerous pests has been the focus as of late for researchers at the University of Trento. The team is relying on the Raspberry Pi to implement a custom machine learning (ML) platform capable of detecting dangerous pests for farmers before the infestation escalates to dangerous levels.
To tackle the issue, they developed solar-powered, pheromone-based smart traps to lure bugs inside. The Raspberry Pi then captures an image for evaluation to determine whether or not too many unwanted bugs are present. In this case, the team has trained their system to detect Codling Moths with a custom model but it would be possible to train the platform to look for other bugs using another model.
According to the paper published by the University of Trento research team, a Raspberry Pi 4 was considered for its notably higher processing power. However, the team ultimately chose to use an undisclosed Raspberry Pi 3 model as it operates with lower average power consumption.
The smart traps are built using the Raspberry Pi, a camera module, a solar power system, a long-range radio chip and an Intel Neural Compute module to help with machine learning. A demonstration of the process identifies Codling Moths against acceptable bugs by highlighting them with a red box. This data is used to track current pest levels and can be used to trigger notifications to farmers when the population gets too high.
This is one of the best Raspberry Pi projects we’ve ever seen for pest control and it’s exciting to see so many Raspberry Pis used on a large scale. To get a closer look at how the system works and what went into its creation, check out the full research paper from the University of Trento.