When a tourist steps off a plane at Praia airport, boards an inter-island ferry, or checks into a resort on Boa Vista, the security of those spaces increasingly depends on technologies invisible to the naked eye. One of them — facial recognition — already protects airports, stadiums, national borders and city centers in over 50 countries. And it’s coming to Cabo Verde.
But what exactly is facial recognition? How does it work in practice? What are its limits and risks? And how can a small island nation benefit from this technology without compromising citizens’ privacy?
At RADAR, we’ve been following this technology closely for years and have recently partnered with Herta — a European company specializing in biometric AI, based in Barcelona, with deployments in more than 50 countries. Here’s what we’ve learned.
What facial recognition actually does
Think about how your brain instantly recognizes someone you know — in a crowded market, in an old photograph, even when they’ve changed their hairstyle. Facial recognition does something similar, but with cameras and artificial intelligence algorithms.
The process works in three fundamental steps:
Detection — The camera identifies that a face exists in the image, separating it from the background. This happens in real time, even in environments with many people in motion.
Mapping — The software analyzes the face’s unique characteristics: the distance between the eyes, the jawline shape, the cheekbone structure. These measurements create a facial signature — a unique mathematical representation, similar to a fingerprint.
Comparison — That signature is compared against an authorized database. If there’s a match, the system generates an alert for a human operator. If not, the image is discarded — the system doesn’t store faces indiscriminately.
This last point is crucial and frequently misunderstood: a well-implemented facial recognition system is not a mass surveillance tool. It’s a verification tool — it compares faces against specific, authorized watchlists such as active arrest warrants, missing persons, or individuals banned from certain locations.
Where it’s already working in the real world
Facial recognition is no longer science fiction. It’s operating in demanding environments worldwide, and the results are measurable.
Public transport in Madrid — One of Europe’s largest long-distance bus stations, handling approximately 20 million passengers per year, implemented facial recognition at entrances and public areas. The result? Security incidents dropped from five per day to just one per month.
Football stadiums in Uruguay — Facing serious match-day violence, Uruguay deployed facial recognition at Montevideo’s Centenario Stadium. The system cross-references fans’ faces against a list of individuals banned from attending matches. The Uruguayan Ministry of Interior rated the security operations as impeccable — zero incidents during monitored events.
Border control in Colombia — To manage the flow of millions of Venezuelan citizens crossing the border, Colombia implemented a biometric platform combining facial recognition, fingerprint and iris scanning. The system registered over two million people, enabling secure identification and legal integration of migrants.
Railways in India — Across 230 train stations in eastern India, cameras with facial recognition monitor public spaces, entrances and exits in real time. On a network carrying millions of passengers daily, the system enables rapid identification of wanted individuals without disrupting normal passenger flow.
The ethical question: security vs. privacy
Any serious discussion about facial recognition must address privacy. And rightly so — this is a powerful technology that, if misused, can become a tool of social control.
The difference between responsible and problematic use rests on three pillars:
Proportionality — The system should only be used where there is a real, documented need. Monitoring an airport entrance is proportional. Surveilling every street in a city is not.
Governance — Who can access the data? Who authorizes the watchlists? How long is data retained? These questions need clear, auditable answers before any camera is turned on.
Human oversight — Facial recognition is a decision-support tool, not an autonomous decision-making tool. The system generates alerts; final decisions must always remain in the hands of authorized operators following documented procedures.
The European Union has led this debate with the AI Act — the world’s first comprehensive regulation for artificial intelligence — which classifies facial recognition in public spaces as a high risk application, requiring transparency, traceability, and mandatory human oversight.
Why this makes sense for an island nation
Cabo Verde has characteristics that make facial recognition particularly interesting — and particularly feasible.
With 10 islands, 7 airports, 9 ports, and nearly a million tourists per year, the country depends on a relatively small number of critical entry and exit points. Protecting those points — airports, ports, high-traffic tourist zones, and government facilities — is more effective and less intrusive than trying to monitor an extensive continental territory.
The compact geography enables something larger countries can rarely achieve: a single platform linking all critical points across islands. Imagine a system where an alert generated at Sal airport is automatically shared with the ports in São Vicente and Praia, using the same protocols, the same rules, the same centralized oversight. This isn’t about surveilling citizens — it’s about coordinating security intelligently across an archipelago where fragmentation is the biggest operational challenge.
Other relevant use cases include post-incident forensic search, locating missing persons in high-traffic tourist areas, and controlling access to sensitive facilities without physical cards that can be lost, stolen, or shared.
European technology as a trust benchmark
In evaluating different solutions, one of the factors that weighed most heavily was the origin and regulatory culture of the technology. European solutions — developed under the scrutiny of GDPR and now the AI Act — tend to build governance, traceability, and human oversight into the system’s architecture itself, not as optional features bolted on afterward.
Herta, the company we’ve partnered with for this area, exemplifies that approach. Founded in Barcelona in 2010, they participated in Spain’s first AI Sandbox and have deployed projects in environments as demanding as international borders, mass transit railways, and high-profile events — always with the principle that every use must be justifiable, auditable, and controllable.
For us as local integrators, that culture of compliance isn’t just a legal requirement — it’s a credibility guarantee before government, businesses, and citizens.
From concept to reality
Facial recognition is not a magic solution and doesn’t replace human presence, security team training, or proper maintenance of existing systems. It’s an additional layer of intelligence that can transform passive cameras (which only record) into active cameras (which alert in real time).
Responsible implementation always begins with an assessment: what are the real critical points? What is the legal basis? Which cameras already exist and can be reused? What operational procedures are needed?
At RADAR, we’re working on those answers — for our clients and for the country. If your facility’s security involves access points, people flow, or the need to identify incidents quickly, this is a conversation worth having.
Schedule a free technical assessment — no obligation. Contact us here.