Skip to content
Esc

Suggestions

Contact · discuss your projectOnline AI assessmentPortfolioAI solutions · hubHow we workPricing · packages and benchmarksFree tools · check a website or a companyInsights · guides and technologiesAI & web glossaryIndustriesWeb design & web developmentFast, clear websites and online stores built to turn visitors into customers.AI agentsAI assistants that follow the steps you set, use only approved tools and complete concrete tasks, with a person making the decisions.LLM & custom AIAI assistants that answer from your company's documents and data, with source references and controlled access.AI voice agentsCalls answered outside working hours too, appointments confirmed and conversations summarised straight into the CRM.Computer visionThe cameras you already have can flag events, count objects and check quality visually.AI automationAutomated workflows between the apps you already use, with AI where it helps and human review where it matters.Finance AI & business intelligenceInvoice data extracted automatically, reports that update themselves and clear figures for decisions.Digital marketing & growthStrategy, content, SEO, paid campaigns and automation connected in one system that brings in clients and can be measured.Custom software & integrationsApplications, platforms and integrations built around your processes, so data flows between systems on its own.Branding & visual identityLogo, colours, typography and a brand guide, so your company looks the same on the website, on materials and on social media.
↑↓ move↵ openEsc close
Let's talk

Computer vision · Drone detection

What flies overheadis seen in time.

Drone detection with computer vision is the use of cameras and AI models to spot drones in the sky, tell them apart from birds or other objects and track them in the frame. We build systems that flag these sightings over an industrial or logistics site or an event and send the alert to the people who decide what happens next; we do not jam or intercept drones.

Let's talkHow it works
DRONE · #071. DRONE · #07AIRCRAFT2. AIRCRAFT
LIVE · CAM 12 · ROOF NE19:26:05

Events

  • 19:26:05New aerial object · NE sector
  • 19:26:06Classified as a drone · tracked
  • 19:26:06Alert to the security team
  • 19:26:30Track saved for the report
Animated illustration · not footage from a delivered project

A small drone easily goes unnoticed

Drones are small and fast, and they usually appear against a changing background: clear sky, clouds, buildings or trees. A guard watching several screens can easily miss one, and ordinary surveillance cameras point at the ground, not at the sky.

  • A drone appears and disappears within moments, often without anyone noticing
  • Existing cameras face gates and aisles, not the airspace above them
  • When someone does see one, there is no clear record of its path, time and duration
  • There is no clear procedure: who is told, what is recorded and when the authorities are contacted

Who it is for

  • Industrial and energy sites that want to know when a drone flies over the area
  • Warehouses, terminals and logistics parks with large perimeters that security staff alone struggle to watch
  • Organisers of outdoor events who need a clear picture of the nearby airspace
  • Security companies that want to add a visual detection layer to the services they offer

Step by step

How the system spots a drone, step by step

The five steps on the animation follow a drone from the first dot in the sky to the alert confirmed by security: the image, detection, classification and tracking, the alert rule and the human decision.

LIVE · CAM 12 · ROOF NE19:26:05

Events

Animated illustration · not footage from a delivered project

Step 1 of 5

The camera watches the sky

Cameras are mounted high, on roofs or poles, and cover the airspace of interest sector by sector rather than the gates. A frame is usable when a drone at the distance that matters takes up enough pixels to have a shape, and the sun or strong lights do not shine straight into the lens. For night-time, thermal cameras are typically added, which see the heat of the motors rather than light.

Configuration options

  • Fixed cameras by sector

    Several fixed cameras cover the sky permanently, sector by sector. This is the simplest option and suits clear perimeters, provided the observation distance stays within what the chosen lenses can handle.

  • Wide camera plus PTZ

    A wide-angle camera detects the object and a PTZ camera turns and zooms in on it, so the operator can see clearly what is flying. It suits large sites where visual confirmation matters more than constant coverage by many cameras.

  • Video correlated with other sensors

    Where radar or radio-frequency sensors from specialist suppliers are in place, their detections are correlated with the video, and the camera confirms what the sensor flagged. This is the option for night-time, long distances or sites with high security requirements.

What you receive at hand-over

  • The detection model adapted to the sky and backgrounds at your site, exported for the edge device or local server in use
  • Cameras, sectors of interest and alert thresholds configured, including the list of authorised drones where there is one
  • Alerts delivered on the agreed channels and, typically, control of the PTZ camera where the camera supports it
  • The incident log, with time, duration, track and video clip for every confirmed sighting
  • The review interface for security officers, the written alert procedure and training for the team
  • A measurement report from authorised test flights at your site, across different light and weather conditions

No strings attached: tell us what the cameras should see and we'll tell you honestly whether it can be done.

Let's talk

What the detection system can do

  • 01

    Spots moving objects in the sky

    The model analyses frames from sky-facing cameras and marks small objects that are moving.

  • 02

    Tells drones apart from birds

    Classification takes into account the object's shape and the way it moves, to separate drones from birds, aircraft or insects close to the lens.

  • 03

    Tracks the path in the frame

    Once detected, the drone is tracked from one frame to the next, and a compatible PTZ camera can be turned towards it automatically.

  • 04

    Alerts the security team

    The alert goes out with a snapshot, the time and the camera that spotted the drone, on the agreed channel: app, email or message.

  • 05

    Keeps a record of incidents

    Each confirmed sighting stays in a log with its time, duration and video clip, useful for reporting.

  • 06

    Combines signals from several sensors

    Where radar or radio-frequency sensors from specialist providers are in place, their detections can be correlated with the video.

How it works

How a drone in the sky becomes a verified alert

The steps marked in yellow are approved by a person.

  • Input
  • AI
  • Action
  • Human in the loop
  • Result
  1. 01

    Input

    Sky-facing cameras

    Fixed, PTZ or thermal cameras cover the airspace above the site.

  2. 02

    AI

    Detection and classification

    The model looks for small moving objects and estimates whether they are drones.

  3. 03

    Action

    Tracking and confirmation

    The path is tracked over several frames before an alert is raised.

  4. 04

    Human in the loop

    Security staff review

    The operator watches the clip and confirms or dismisses the alert.

  5. 05

    Action

    Agreed procedure

    Confirmed incidents are logged and passed to the designated people.

  6. 06

    Result

    Incident log

    You have a record of sightings, with time, duration and video.

What drone detection needs

  • Cameras facing the airspace of interest, with a resolution and lens suited to the observation distance
  • Real footage from the site in different weather and light, including birds and other objects that can be mistaken for drones
  • Footage of drones of several types and sizes, filmed lawfully, for training and testing
  • A map of the area: the perimeter, the directions of interest and the points where cameras can be mounted
  • A written procedure for alerts: who receives them, what they check and whom they notify
  • Processing resources close to the cameras, usually an edge device or a local server with a GPU

Possible integrations

  • IP cameras with an RTSP stream
  • ONVIF-compatible PTZ cameras
  • Thermal cameras
  • NVRs and VMS systems
  • Radar or RF sensors from specialist providers, via their API
  • NVIDIA Jetson edge devices
  • Local servers with GPUs
  • REST API and webhooks
  • MQTT
  • Notifications by email, SMS or Microsoft Teams
  • IP cameras with an RTSP stream
  • ONVIF-compatible PTZ cameras
  • Thermal cameras
  • NVRs and VMS systems
  • Radar or RF sensors from specialist providers, via their API
  • NVIDIA Jetson edge devices
  • Local servers with GPUs
  • REST API and webhooks
  • MQTT
  • Notifications by email, SMS or Microsoft Teams

Each integration's compatibility is checked during the analysis phase, for the systems and versions you use.

Control and transparency

What stays with people

  • Confirming each alert from the video clip before any further action
  • Deciding whether to notify the competent authorities and what information to pass on
  • Setting the areas of interest and the alert thresholds, so that not every nearby flight is flagged
  • Regularly reviewing false alerts and missed drones to fine-tune the system

Limits and operating conditions

  • The system only detects and alerts; jamming, intercepting or taking control of a drone is reserved for authorised institutions and is not part of what we build
  • The distance at which a drone can be spotted depends on its size, the camera and the lens; a very small or very distant drone may not be visible at all
  • Fog, rain, the sun in the frame or busy backgrounds reduce detection quality, and birds can be mistaken for drones
  • The image shows the drone, not the person flying it; the system does not identify the operator and does not track people

How we evaluate drone detection

We agree the indicators together before launch and track them afterwards. The values depend on the process, the data and the volume.

Detection rate
Planned, authorised test flights in the area, compared with the alerts the system raises
False alerts
Alerts dismissed by operators, counted per week and grouped by cause: birds, aircraft, reflections
Time to alert
The interval between the drone appearing in the frame and the alert, measured on test recordings
Behaviour in different weather
Results compared separately on clear, overcast and rainy days and at night

How we build a drone detection system, in 4 steps

  1. 01

    We assess the site and the legal framework

    We agree the areas of interest, the camera positions and what you may lawfully do after a detection, before anything is installed.

  2. 02

    We test cameras and collect footage

    We mount trial cameras, record real conditions and organise authorised test flights for training data.

  3. 03

    We train and calibrate the alerts

    We adapt the model to your sky and backgrounds and tune the thresholds so that alerts are rare but relevant.

  4. 04

    We integrate and monitor

    We connect the alerts to your security procedure, then monitor false alerts and misses and retrain when needed.

Use cases

The scenarios below are conceptual examples, designed to illustrate realistic applications. They are not implementations delivered by WEB-AZ.

  • Energy

    Overflights above a substation

    Sky-facing cameras flag drones that appear over the perimeter, and the security dispatcher confirms the alert and follows the internal procedure.

    Conceptual example
  • Logistics

    The perimeter of a logistics park

    A PTZ camera turns automatically towards the detected drone, so the guard can see clearly what is flying and log the incident.

    Conceptual example
  • Events

    The airspace over a festival

    During the event, the security team receives alerts with a snapshot and direction, and the organiser's authorised drones are marked separately.

    Conceptual example

Useful guides

Before you decide, read

Frequently asked questions

  1. What is drone detection with computer vision?

    It is the use of cameras and AI models to spot drones in the sky, tell them apart from birds or aircraft and track them in the frame. The result is an alert with a snapshot, time and direction, sent to the people who decide what happens next.

  2. Can you stop or jam a drone?

    No. Jamming, intercepting or taking control of a drone is generally reserved for institutions authorised by law, and a private company is not allowed to do it. The systems we build only detect and alert; the next step is your internal procedure and, where appropriate, notifying the authorities.

  3. From what distance can a drone be spotted?

    It depends on the size of the drone, the camera resolution, the lens and the weather. That is why we do not promise a fixed distance but measure it at your site with test flights, before deciding how many cameras are needed and where.

  4. Does it work at night?

    Ordinary cameras struggle to see a drone against a dark sky, especially if it has no lights. Night-time monitoring usually relies on thermal cameras or additional sensors, and the system's behaviour is tested separately in those conditions.

  5. Does the system identify the person flying the drone?

    No. Sky-facing cameras see the drone, not the operator, and we do not do facial recognition or identify people. The information in the incident log can be made available to the authorities, who have the power to investigate.

  6. What legal obligations should be considered?

    Drone use is regulated at EU and national level, and in Romania the competent body is AACR (the Romanian Civil Aeronautical Authority). The cameras you install must also comply with data protection and video surveillance rules. We review together what the cameras cover and how recordings are kept, and for specific legal questions we recommend a specialist adviser.

Can’t find the answer? Write to us and we’ll reply directly.

Ask a question

Want to know when a drone appears over your site?

Describe your site, its perimeter and the cameras you have. We will tell you what can realistically be detected and what should be tested on site.

Prefer to talk directly?

Message us on WhatsApp+40 736 469 828

Email: office@webaz.ro

Project type *
Add details, budget and timeline (optional)

* required fields. We don't send newsletters.

Related solutions