FaceAir — Drone Face Recognition
Drone face recognition with FaceAir brings real-time identification to the sky, turning live drone video into actionable intelligence across areas too wide for fixed cameras.
Request a demo
FaceAir by Alfabeta is a drone face recognition system that mounts our recognition engine on aerial platforms and identifies and tracks individuals in real time from the air. It is built as aerial surveillance AI for scenarios where fixed cameras cannot reach — large events, long perimeters and search operations spanning many hectares of open ground. The system detects and locates faces in live footage even in complex, moving scenes, giving security and response teams a vantage point that ground cameras simply cannot provide.
As UAV video analytics, FaceAir ingests the drone's stream over the RTSP protocol, USB, or effectively any streaming protocol, and runs deep-learning recognition against it. The algorithms are tuned for high precision to minimise false positives, and can detect and track multiple faces simultaneously within the same frame. Effective recognition holds across varying altitude and range for as long as facial attributes remain visible to the camera, so operators can trade coverage area against identification confidence by adjusting how the drone is flown.
Detected faces are matched against authorised watchlists, and users can configure custom alerts that fire the moment a recognised individual appears — a capability that strengthens active security protocols rather than only recording after the fact. All recognition data is logged and timestamped for later analysis and record-keeping, and an API is available so FaceAir can be integrated and customised inside a wider command, CCTV or ground-camera network for a single continuous view of a target across a site.
The platform is designed with compliance and data privacy in mind: it complies with data privacy regulations and lets operators configure their own data retention policies, so aerial identification can be deployed responsibly. Optional on-device inference keeps the system identifying subjects even when the link back from the drone is weak or intermittent, which matters in the field conditions where aerial coverage is most needed.
Aerial face detection
Detects and tracks faces from moving drone footage across wide, open areas, extending coverage far beyond fixed cameras. It works in complex environments and can follow multiple faces in the same frame at once.
Watchlist matching
Matches faces in-flight against persons-of-interest lists using high-precision deep-learning algorithms tuned to minimise false positives. Custom alerts fire the moment a flagged subject is identified.
Edge inference
Runs recognition on the edge to keep identifying subjects even when the network link to the drone is weak or intermittent. On-device processing keeps identification responsive in the remote field conditions where aerial coverage matters most.
Crowd security
Scans dense crowds for flagged individuals and hazards, supporting proactive response at large public gatherings. Detecting and tracking many faces simultaneously lets a single drone cover ground that would need dozens of fixed cameras.
Ground-camera fusion
Correlates aerial detections with fixed CCTV networks through an available API for a single, continuous view of a target across the site. Aerial and ground feeds combine into one operating picture rather than isolated streams.
Live command view
Streams positions, tracks and alerts to a central dashboard so operators can direct teams in real time. Every detection is logged and timestamped for analysis and record-keeping after the operation.
Frequently asked questions
What video sources and streaming protocols does FaceAir support?
FaceAir is designed to accept the drone's live feed over the RTSP protocol, USB, or effectively any streaming protocol the platform provides. That flexibility lets it slot onto a range of drone hardware and integrate with existing streaming infrastructure rather than requiring a specific proprietary camera.
At what altitude and range can it recognise a face?
Recognition is effective across varying altitude and distance, for as long as the facial attributes remain visible to the camera. In practice that means altitude and range are a trade-off against identification confidence: fly lower or use greater optical detail for reliable matching, or fly higher to prioritise area coverage. There is no single fixed number because it depends on the camera, optics and how the drone is flown.
Can it track more than one person at a time?
Yes. The system can detect and track multiple faces simultaneously within the same frame, which is what makes it suitable for crowd security at events where many people are in view at once. Recognised subjects can each trigger their own configurable alert.
How does FaceAir handle data privacy and compliance?
The system is built to comply with data privacy regulations and lets operators configure their own data retention policies. All recognition data is logged and timestamped for accountability, and retention can be tuned so aerial identification is deployed responsibly within the operator's legal framework.
Can it be integrated with our existing CCTV or command centre?
Yes. An API is available for customisation and integration, so aerial detections can be correlated with fixed ground-camera networks and streamed into a central command dashboard. The aim is a single continuous view of a target across the whole site rather than a standalone drone feed.