Nigeria is losing its forests at a staggering pace. Once home to about 20 million hectares of forest, the country has lost nearly 96% of its primary cover, according to Global Forest Watch. From 2001 to 2024, Nigeria lost 1.44 million hectares (Mha) of tree cover, equivalent to a 14% decrease since 2000, driven by logging, agriculture, and weak enforcement.
Despite policies like the Reducing Emissions from Deforestation and Forest Degradation (REDD+) strategy adopted by the federal government aimed at reducing deforestation, and Nigeria’s climate commitments under the Paris Agreement, illegal logging continues especially in under-policed reserves across states like Edo, Cross River, Ogun, and Rivers. With no real-time monitoring tools or reliable forest patrols, most of it goes unnoticed.
In the heart of this crisis, three 18-year-old teenagers from Port Harcourt—Lesley John Jumbo, Bright Sunday, and Blessed Pepple—are engineering what they hope will spark some change. Their invention, Reforest AI, is a low-cost forest-monitoring system that uses artificial intelligence, embedded systems, and IoT (Internet of Things) sensors to detect and deter illegal logging—even in areas without power or internet.
Their motivation wasn’t born in a lab or a climate conference. It began with a simple, hard-to-ignore question: why do we keep watching forests disappear without doing anything?
Earlier this year, their innovation earned Top Honours at the 2025 Slingshot Challenge, a global competition hosted by the National Geographic Society and the Paul G. Allen Family Foundation. Selected from more than 2,700 entries across 96 countries, the Nigerian team received a $10,000 grant to further develop and deploy their solution.
From hackathon to hardware
The trio first teamed up at a Technoville Nigeria hackathon in Port Harcourt, the rivers state capital. John Jumbo and Sunday were classmates; Pepple was John Jumbo’s neighbour. What united them, the team explains, was “this shared drive to build cool, meaningful stuff with tech.”
The turning point came at another hackathon, hosted by Techrity in Port Harcourt, when their conversations turned to the forests around them. “We realised that if we just keep watching environmental damage happen without doing anything, we’re part of the problem,” the group explained. That event gave them the urgency to turn a vague worry into a real solution.
Instead of launching awareness campaigns, they decided to build a tool that could address a specific aspect of the problem they’d pinpointed. Research drew their attention particularly to illegal logging as a silent but devastating contributor to climate change in Nigeria. Forests like the Cross River Reserve, once part of a vast West African canopy, now exist in fragmented national parks. The Niger Delta’s mangroves are shrinking fast, degraded by both chainsaws and crude oil.
So they got to work.
Each member brings a complementary skill to the group: John Jumbo, an embedded systems engineer, focuses on the software and leads outreach and storytelling around the project. Sunday, also an embedded systems engineer, handles all things electronics. Pepple, a designer, ensures everything looks great and works intuitively.
In a bold move to demonstrate their commitment, they paused their education to build Reforest AI full-time. “Where we’re from, climate action isn’t really taken seriously, so trying to build a solution for a problem most people don’t even recognise felt almost impossible. But that just made us double down,” John Jumbo explains.
“Formal education didn’t really play a big role in what we built,” the team notes. “Everything—from coding to hardware—we taught ourselves.”
Their first fund came from local believers: family, mentors, and fellow builders who had seen their work. That early support helped them take the idea beyond prototype.
Reforest AI: A smarter, scalable forest guard
The initial idea for Reforest AI was to strap individual sensors to trees using accelerometers and gyroscopes to monitor changes in their angles—so if someone was trying to cut one down, they’d know. But they quickly realised the obvious problem. “If there are a billion trees, are we really going to deploy a billion devices?” John Jumbo says with a laugh. “Not scalable.”
So they pivoted. Instead of monitoring individual trees, they created smart sensor towers that form a mesh network. These sleek, minimal towers act as smart nodes in a distributed network, using GPS to triangulate positions and cover large areas efficiently.
The breakthrough came when they made the system AI-powered. They trained a sound recognition model that can detect chainsaws, axes—basically the audio signature of illegal logging activities.
“We trained the system to listen like a forest ranger who’s been out there for years,” Sunday explains. “It doesn’t freak out over every sound—just the ones that matter.”
When a threat is detected, the system doesn’t just sit quietly—it responds. First, it triggers a deterrent: ultra-bright lights and pre-recorded audio warnings to scare off intruders. Then it activates its onboard cameras and microphones to stream live footage to authorised personnel—forest rangers, local authorities, whoever’s in charge. That way, they can instantly assess the situation and intervene if needed.
Crucially, it doesn’t need the internet to function. The system uses local mesh protocols like ESP-NOW (a wireless communication protocol developed by Espressif, enabling direct, low-power communication between devices without the need for a router) to share data instantly between devices—even in remote areas like Edo State’s 1,000-square-kilometre Okomu Forest Reserve.
“If the tech needs internet to work, then it’s not built for the people who need it most,” Sunday stressed.
The innovation caught the attention of the National Geographic Society. “These young innovators are not only identifying urgent environmental issues in their own communities, but they’re also developing tangible, thoughtful solutions,” said Deborah Grayson, the Society’s Chief Education Officer, in a press release.
Engineering for extreme environments
Building technology for Nigeria’s tropical forests meant overcoming brutal conditions: humidity exceeding 80%, temperatures reaching 35°C, and annual rainfall surpassing 3,000mm in some areas.
“Making sensors that survive in tropical forests was no joke,” John Jumbo says. “The weather is insane—crazy humidity, nonstop rain, and heat that just won’t quit.” Waterproofing and making the devices rugged became essential.
Power proved equally challenging with no electrical infrastructure. Their solution combines solar panels with intelligent power management, using smart sleep cycles that only wake the system when it hears a potential threat. They optimised for hardware that balances performance with low energy usage, doing most of the heavy lifting on-device instead of relying on the cloud.
The most complex challenge was audio recognition in naturally noisy environments. Nigerian forests teem with over 4,700 plant species and hundreds of animal species creating constant soundscapes.
“Forests are noisy—birds, rain, wind, insects,” they note. “Teaching our system to tell ‘chainsaw’ from ‘just a noisy bird’ took a lot of tweaking and patience.”
They fed their AI extensive audio data—chainsaws, axe hits, and human movement balanced against wind, birds, rain, and wildlife calls including monkeys. The system learned to distinguish between nature and human threats.
“We’d rather it give a heads-up that gets checked out than stay silent when a tree’s going down,” they said.
For the technical implementation, they leaned heavily on open-source tools. TensorFlow Lite handled the AI components, running sound recognition on the edge without draining power. PlatformIO made firmware development smoother, while libraries like TinyML and Edge Impulse helped train and deploy models quickly.
Real world limitations
The team maintains honest assessments of their system’s constraints. Sound interference remains significant—storms can mask chainsaw activity, and sophisticated loggers might use quieter tools or traditional methods that produce less noise. Dense forest canopies can disrupt mesh communication between nodes, and different forest terrains may require system adjustments.
“We’d rather ship and learn than wait for perfect,” John Jumbo explains. “Every failure teaches us what to fix next.”
They track several key metrics: detection accuracy (how well the system picks up real threats without false alarms), response times (how fast it goes from collecting data to alerting rangers), intervention success rates (how often someone actually steps in because of a Reforest AI alert), and long-term tree loss changes in monitored areas—which serves as their north star metric.
Feedback from rangers and community leaders provides ground-truth validation. “If they say, ‘this helps us do our job better,’ that’s a win for us,” the team explains.
Growing interest and focused vision
Despite being early-stage, Reforest AI has attracted environmental agencies and NGOs who are keen on exploring pilot programmes. Most encouraging, according to the team, has been community leader support, even from those who don’t fully grasp the technology.
“Some of them don’t even fully get the tech yet, but they get why it matters,” Sunday explains. “They’ve told us, ‘If this helps us protect our land, let’s figure it out together.’ That’s brilliant.”
The innovation has also garnered support from those working on the front lines of forest protection.
Patrick Emaiku, a forest guard in Benue State, welcomed the development enthusiastically. “This is a welcome development that will really help us protect the forest more,” he told TechCabal. Emaiku explained that Nigerian forest security currently uses drone technology to monitor forests, but noted that drones are not being deployed across all states.
However, he highlighted a critical challenge that could affect the innovation’s deployment. “We have a bill in the Nigerian Senate awaiting presidential assent. If that bill is not signed into law, the real potential of this innovation might not be realised,” he explained, referring to the Nigerian Forest Security Service (Establishment) Bill, 2025, which was recently passed by the National Assembly and is currently awaiting presidential approval.
The bill aims to tackle insecurity around the country’s forest reserves. “Every forest in the country is under the control of bandits and kidnappers who use them as hideouts to carry out attacks,” Emaiku said, explaining that without the bill’s passage, the forest security service lacks sufficient backing and resources to properly deploy and utilise innovations like Reforest AI.
These insights from both community leaders and forest professionals underscore the complex ecosystem that innovative solutions like Reforest AI must navigate—where technical capability meets policy implementation and community acceptance.
While expanding capabilities might seem logical—wildlife tracking, fire detection, and other features—the team remains laser-focused on illegal logging. “We don’t want to half-solve ten problems,” they explain. “We want to fully solve one. Logging is a major driver of deforestation, and if we can build something that actually makes it harder to illegally cut down trees, that’s already a big win.”
Scaling Reforest AI across Africa
Their expansion strategy prioritises community partnerships over top-down deployment. Within Nigeria, they plan state-level pilots with forest agencies, conservation organisations, and even local vigilante groups. The goal is to show that the system works, refine it with feedback, and create real case studies that other regions can’t ignore.
Across Africa, they envision a community-first model: partnering with grassroots organisations already doing the work and training local builders to maintain and customise systems for regional needs.
The approach aligns with Nigeria’s Paris Agreement commitments. Forest protection plays a crucial role in the country’s pledge to reduce greenhouse gas emissions by 20% unconditionally—or 45% with international support—by 2030.
“Nigeria has made big commitments under the Paris Agreement,” the team notes. “But the gap between goals and execution is still real. That’s where Reforest AI fits in. If you can detect illegal logging in real time, you’re not just talking about climate goals, you’re enforcing them.”
A blueprint for youth-driven climate innovation
To other young Africans thinking about climate tech, the team offers a message grounded in action over perfection.
“If you’ve been thinking about it for so long, this is probably the best time to act,” John Jumbo stresses. “Don’t wait for perfect—no one ever gets it.”
Their story is a blueprint: identify a local problem, teach yourself the tools, and build something real. The project initially took them about a week and a few days to build, though they’re constantly improving its capabilities. What made the difference was having people who saw the vision early on—siblings, mentors, and colleagues who stepped up with support when the team was building from scratch.
Reforest AI will be showcased at the upcoming National Geographic Explorers Festival in June, where the team will join fellow innovators and conservation leaders from around the world. With increasing interest from climate-focused organisations and African innovation accelerators, they’re now exploring partnerships, funding, and mentorship opportunities to commercialise their technology and integrate it into broader environmental protection efforts across the continent.
Reforest AI is more than a forest-monitoring system—it’s a reminder that Africa’s biggest challenges will be solved not just by policy, but by people who dare to build.
