Satellite AI camera trap protects rainforest elephants from poachers

Protecting endangered wildlife in dense rainforests has always been an uphill battle.

Traditional camera traps store footage on SD cards that rangers must hike in to retrieve weeks later. By the time someone spots a poacher or a herd of elephants on the memory card, the damage has long been done.

Researchers at the University of Stirling and tech start-up Hack the Planet just built an AI camera trap that talks directly to satellites. It works deep in the jungle without Wi-Fi, cell service, or long-range radio.

Edge AI meets satellite uplink in Gabon

Instead of beaming heavy raw footage over expensive satellite connections, the device runs an on-camera AI model. The system was trained on thousands of images of animals and people to automatically classify what walks past the lens.

When a threat or an elephant is spotted, the camera tags the image and pings rangers in real time via satellite.

The team tested five camera traps over 72 days in the remote rainforests of Gabon, publishing their findings in Methods in Ecology and Evolution. During the trial, the devices snapped over 800 photos, including 217 of elephants.

The onboard model clocked an 82% accuracy rate in recognizing elephants. More importantly, rangers received critical alerts in an average of seven minutes.

Stopping poachers and saving farmers

This pilot is about more than catching illegal hunters.

According to The Times of London, Gabon is home to around 95,000 critically endangered African forest elephants, roughly 70 percent of the world population.

Their raids on local crops have sparked intense backlash from farmers, who end up killing around 50 elephants each year in self-defense or revenge. A seven-minute heads-up gives villagers and rangers a chance to deter the animals before crops are destroyed and shots are fired.

Why it matters

We think the real brilliance here is treating bandwidth as the ultimate constraint. Satellite internet is brutally expensive and power-hungry; sending full photos from the middle of nowhere simply does not scale for conservation budgets.

By executing inference on the edge and only transmitting lightweight alert packets through satellites, the team cracked the latency problem without blowing up battery life or data costs. It turns passive environmental monitoring into an active intervention tool.

That said, an 82% accuracy rate and a seven-minute turnaround leave room for error. A lot can happen in seven minutes when a multi-ton elephant is heading toward a farm, and false alarms will quickly burn out understaffed ranger units. But as low-power edge chips and satellite messaging become cheaper, this blueprint is likely to become the standard for anti-poaching and forest protection worldwide.