Researchers at the University of South Florida are turning to artificial intelligence and satellite imagery to track down mosquito breeding grounds across Hillsborough County, as local health officials grapple with a dengue fever outbreak. The strategy pairs algorithms with aerial imagery to pinpoint standing water in tires, birdbaths and flowerpots before crews move in to treat it, often the same day the habitat is found.
A Smarter Hunt for Standing Water
USF Associate Professor Benjamin Jacob leads the effort through the university’s Department of Biostatistics and Data Analysis, working to identify and eliminate mosquito breeding grounds as dengue fever cases rise in the Tampa Bay area, according to FOX 13 Tampa Bay. The researchers use satellite imagery to narrow in on potential mosquito habitats to within about 10 meters, then apply geographic mapping to direct treatment crews to exact locations rather than relying on broad aerial or ground spraying. Jacob has described the approach simply: find the breeding sites and destroy them, instead of treating everything with insecticide.
The underlying technology has roots that go back several years. Jacob previously created a smartphone app that pairs his algorithm with drone and satellite images to identify previously unknown mosquito breeding habitats so they can be treated within the same day, according to USF. That mapping work once revealed more than 9,000 mosquito habitats carrying dengue and Zika viruses across Hillsborough, Manatee and Polk counties, and testing found the drone-based system was 100% accurate in locating bodies of water where mosquitoes were most likely to breed.
From Tampa Bay to Africa and Southeast Asia
Jacob’s work extends well beyond Florida. He launched a program called Seek and Destroy that trains government agencies in Cambodia, Uganda, Kenya and Rwanda to use the app in areas hit by infectious disease, the university’s report notes. Jacob, who also works in Africa, has said that identifying smaller breeding sites is critical to controlling mosquito populations, and he calls the overall team strategy Seek and Destroy — the same name he gave the international training program.
USF’s broader research on this front includes separate work by professor Sriram Chellappan and student assistant Mona Minakshi, who developed an AI technique that detects mosquito breeding habitats in drone video and extracts their GPS coordinates, per the university’s engineering news page. Those frames and locations are sent back in real time to public health officials so crews can act on them quickly. During a field test in Manatee County, Jacob and his assistants verified that habitats flagged by the computer algorithms did in fact contain mosquito larvae.
Targeting Instead of Spraying Everything
The goal of narrowing in on specific habitats, rather than spraying entire fields, is to cut down on overall insecticide use, according to the same USF materials. Researchers have also proposed letting drones detect a breeding ground and then spray a controlled amount of insecticide directly onto it rather than broadly dousing large areas — though a planned Manatee County test of that insecticide-delivery system was delayed by the COVID-19 outbreak, and the evidence does not show whether it was ultimately carried out…