Every morning before sunrise, Sagar Tamang heads to his maize fields in Birta Deurali village, scanning the treeline for movement. The 46-year-old farmer isn't alone — for years, he and his neighbors have been locked in an exhausting battle with Nepal's rhesus macaques, a battle they often seem to be losing. "If you leave even for a short time, the monkeys do considerable damage," Tamang said.
In the central Nepal district of Kavrepalanchok, monkeys have become a fact of life that costs farmers real money. Sunmaya Lama, 32, lost roughly $670 worth of crops over the past three years. When she asked local officials for help, she was told there was no compensation program. "So we just bear the loss ourselves," she said.
The problem extends across the country. A 2026 study found that nearly half of a macaque troop's diet in central Nepal now comes from cultivated crops. The monkeys can damage maize from the moment a seed sprouts until it reaches storage. About 44% of Nepal's land provides suitable macaque habitat, yet less than 8% falls within protected parks.
Now, help may be coming from an unexpected source: artificial intelligence.
Progress Jung Thapa, a researcher at Madan Bhandari University of Science and Technology, has spent the past two years building a system that uses cameras and AI to detect monkeys in real time and figure out what they're doing. The technology works like a security camera that never blinks, checking every frame and asking, "Is that a monkey? And what is it doing?"
The system learned to recognize monkeys the way a child learns to recognize a dog — by being shown thousands of photos first. Thapa and his team trained it on more than 4,000 annotated images, and it can now distinguish between a monkey that's just watching and one that's already eating. That distinction matters: a monkey that's watching gives a farmer a few extra seconds to respond, while a monkey already feeding needs immediate action.
In 28 field tests, the system achieved around 88% accuracy. The underlying detection model performs even better, correctly identifying monkeys roughly 92% of the time while running smoothly on affordable hardware.
"The monkey mind is often described as restless," Thapa said. "But what I observed is the opposite. They are extremely present." He sees the technology not as a replacement for human judgment, but as a tool that gives farmers an edge. "A machine can detect. But a human decides."
The system isn't ready for widespread use yet — Thapa and his team are still refining it for the unpredictable conditions of real farms. But in a country where the government has yet to report on a task force formed to address the conflict, researchers like Thapa are building something practical: a way for smallholder farmers to fight back, one detected monkey at a time.