In 2021, deep in the Carpathian Mountains of Romania, a small forest protection NGO working to identify and persecute perpetrators of illegal logging began receiving alerts it couldn’t quite believe.
The notifications weren’t from rangers or local informants; they were coming from a computer. A system quietly cross-referenced video feeds from remote checkpoints, measured the volume of timber in trucks, checked license plates against the national logging registry, and flagged suspicious loads.
The software had been designed by Teodor Calin, or Teo for short, then a university student who’d spent most of his life teaching machines how to perceive the world. What began as a side project soon evolved into a functioning deterrent for environmental crime. “At first, I just wanted to see if a computer could measure what a human eye might miss,” Calin says. “But once the system started catching real cases, I realized it wasn’t about code anymore, it was about responsibility.”
Over the next four years, that system helped identify roughly 400 incidents of illegal logging, representing around 8,000 cubic meters, or 6,000 tonnes, of stolen timber. The fines (totaling nearly $450,000) funded the NGO’s continued operations and reshaped Calin’s understanding of what artificial intelligence could achieve in the physical world.
Illegal Logging: An Invisible Crime
Romania’s forests are among the most biologically rich in Europe, home to bears, lynx, and ancient oak and beech trees. But they have also been targets of illegal logging for decades.
Much of the stolen timber moves quickly, cut in remote areas, loaded onto trucks, and driven to mills with forged paperwork. By the time inspectors can respond, the evidence is often gone.
Calin’s idea was simple: use computer vision to cross-check what human inspectors couldn’t. “Every truck has a licence plate,” he explains. “Every legal transport has a registered volume of wood. If the truck is carrying more than the documents say, that’s evidence.”
His program analyzes the shape and volume of logs in each image and calculates how much timber is on board. It then compares that figure to a national database of authorised transports. If the numbers don’t match, it automatically raises a red flag.
Four Years, Hundreds of Cases
The system went live in 2021 as a pilot for a small forest protection NGO. Within months, it began identifying suspicious trucks almost daily.
Over the next four years, the camera network recorded more than 400 separate cases of unauthorised hauling, about 8,000 cubic metres of wood, or nearly 6,000 tonnes of timber. Authorities traced and fined many of the operators, resulting in an estimated $450,000 in penalties. Those funds now help finance the NGO’s field patrols and educational campaigns.
“It was the first time we could verify transports without chasing trucks down the road,” says one ranger involved in the project. “The software gave us proof in seconds.”
The system’s success has drawn attention from conservation groups and local officials. Several counties have requested similar installations, and the NGO now manages a small network of cameras across different transport routes.
A Model for Other Regions
Illegal logging isn’t unique to Romania. Across Eastern Europe, Latin America, and Africa, similar challenges exist: long forest borders, limited manpower, and complex supply chains.
Calin’s system has inspired pilot projects in neighbouring countries, where authorities are experimenting with using vision-based monitoring to flag irregular shipments.
Environmental experts see potential beyond forestry. The same kind of measurement and verification could support anti-poaching efforts, waste tracking, or even construction-site monitoring. “Anywhere there’s movement of resources that should be accounted for,” says one environmental analyst, “this kind of technology can help.”
From Forests to Factories
The success of the forest surveillance system became the seed for Calin’s next venture, Vulture Labs. Drawing on the same principle of teaching cameras to understand what they see, his team began adapting the technology for new environments.
Now, the same algorithms that once tracked timber trucks are powering spatial intelligence tools for retail, manufacturing, and logistics, turning ordinary cameras into proactive sensors that detect risks, optimize workflows, and improve safety.
“We realized the problem wasn’t just in the forests,” Calin says. “It’s everywhere, people rely on visual data but lack the means to interpret it. We’re helping cameras become partners, not just observers.”
Teo’s Cameras Are Still Watching
On a cold morning in Transylvania, one of the original cameras still records every truck that passes. Its data feeds into a small control room where alerts light up on a simple dashboard. When a red notification appears, it means something doesn’t add up.
For Calin, those small alerts represent far more than numbers on a screen. “Each one is a piece of forest that stays where it belongs,” he says. “That’s what matters.”
Three years after its launch, his system continues to run quietly. And in a region where illegal logging once seemed unstoppable, that’s proof that sometimes technology can do what bureaucracy and manpower cannot: see everything, and forget nothing.
