Russian AI Scans Satellites to Hunt Toxic Hogweed
Neural networks trained on 30 years of satellite data can now identify hogweed infestations with 87 percent accuracy—and even predict where the invasive plant will spread next.
Russian scientists have taught artificial intelligence to spot poisonous plants from orbit. A team in Perm built a system that detects hogweed clusters with striking accuracy—87 percent, according to its creators.
Fifteen cartographers helped train the model. They fed it three decades of archived satellite data and tuned it to read multiple spectral channels. One year later, the neural network was ready—and it works.
The algorithm goes beyond mapping today’s infestations. It predicts new outbreaks, flags high-risk areas, and can even catch the first sprouts. It also measures how effective treatment efforts have been. Armed with this data, officials can plan smarter eradication campaigns and limit the damage hogweed inflicts.
The threat is real. Hogweed spreads fast, choking out native plants and destabilizing ecosystems. For people, it’s even more dangerous: a brush against its leaves can cause painful chemical burns.
Russia’s breakthrough could matter far beyond its borders. Hogweed plagues Europe, North America, and Asia too. With AI now on the case, the fight against one of the world’s most invasive plants might finally gain ground.
