Face recognition, weapon/knife detection, perimeter protection and crowd analysis in a single panel. A security infrastructure that sends no footage to the internet and runs entirely on your local network.
Cameras, the analysis engine, the built-in AI and the database all run on the same local network, as a fully closed, inline system. It is designed to work at full capacity even without an internet connection.
IP cameras connect through a PoE switch directly to the server that runs the system. Footage is never sent to external servers at any stage; the system runs fully inline and closed to the outside.
Face recognition and threat detection run on the local GPU in milliseconds, with no cloud API latency and no risk of outages.
Functions such as natural-language log search and incident summaries are powered by an AI model that runs entirely on site. No query or data is sent outside, not even for model inference.
Every event is written to a hash-chained audit trail and sensitive data is protected with strong encryption. If the database is altered from outside, the chain breaks and the system detects it immediately.
Face recognition signatures and photos of registered people are stored encrypted in the database with a separate key. Even if the database file alone is captured, no biometric data can be read without the encryption key.
Licensing and software updates are verified with an Ed25519 digital signature, and the private key is never included in the installation. An unsigned or forged update package is never applied to the system.
Each module can be switched on or off independently and has its own cameras and rules. Run only what you need.
🚪
Recognises every person passing through a door in real time, classifies them as authorised, unauthorised or unknown, and creates a record with a photo. Each camera can have its own identity list and rules, and multiple entry points are managed from the same panel.
🛡️
Monitors open areas, warehouses and restricted sites with zone-based authorisation and reports at once when someone without access enters. Each zone carries its own authorised list, so a person can be allowed in one zone and unauthorised in another.
🔪
The moment a weapon or knife appears in the image, the system automatically puts that camera into "high-risk area" mode: it increases the scan rate and lowers the full-alarm threshold so a second sighting is caught much faster. A silent evidence record is saved even at the first sighting, so no scene disappears without a trace, even if nobody stays in front of the camera for more than a few seconds. It runs on a locally trained, continuously improved image recognition model.
👥
Sample view
Measures the live density of people in an area and alerts you when your threshold is exceeded, providing density management for event areas, entry points and shared spaces.
Note: The image is a sample illustration of how the feature works; it is not a real scene or measurement data. Counting accuracy varies with camera angle, lighting and crowd density.
🏢
Recognises every participant in the meeting room automatically as soon as the camera starts, separates authorised from unauthorised people and builds a live attendance list, with no need for a manual sign-in sheet, QR code or hand-entered attendance. When an unauthorised or uninvited person enters the room, the system notices at once and notifies the relevant team; participants who hide their face from the camera are flagged separately. Each meeting is recorded with its own start and end timestamps and can be compared with past attendance data.
🎥
Automatically records the moments an alarm is triggered. It creates directly viewable evidence clips that include the seconds before and after the event and links them to the alarm history.
🔍
Queries the passage and alarm history of people registered in the system's own database: through which door and when a person passed, and in which zones they were seen. No image is sent outside and no web or social media search is performed.
📊
Automatically produces daily, weekly and monthly summaries every night, every Monday and at the start of every month, with exact figures and a tamper-proof audit trail. Monthly reports adjust to the calendar: on the 23rd of the month they cover days 1-23, and on the last day they cover days 1-30/31.
🔌
Create a key to control the system from outside, and when an alarm is triggered send an instant notification to your own server, n8n or Zapier. Automate actions such as unlocking a door, sending an SMS or saving to Excel.
When an alarm is triggered, the notification is not one-way. The manager responds directly in Telegram with the buttons under the incident photo: unlock or lock the door remotely, confirm the alarm or dismiss it as a false alarm. You respond at the moment of the event, from your phone with a single tap, without ever opening the panel.
You do not rewind recordings to find out "what happened". The second a threat appears on camera, it is on your phone. Vision X Future cuts the time between detection and notification to almost zero, so you can act while the incident is still going on, not after it has ended.
The faces and photos of registered people are not kept in the database as plain text but encrypted with a separate key. Even if the database file alone is obtained, whether copied, stolen or leaked, none of the biometric data inside can be read without the encryption key.
Go beyond recognition; bring the decision to the door and the approval to your phone. Three new features, all working by your rules.
🚪
When an authorised person arrives the door opens by itself; if an unauthorised person is beside them it stays shut. You decide which status opens which door.
Note: The image is an illustration. Relay and lock hardware is optional; VXF only sends the command, device choice and physical installation belong to the customer.
📲
When an alarm arrives, the administrator picks one of the predefined actions in Telegram; it runs only after a second confirmation and can be undone.
Note: The action name in the image is an example; the customer defines actions and connected devices.
🧑
A clear face photo of everyone passing is stored separately; unregistered people can later be found with Past Face Search.
Note: The image is an illustration; processing of personal data is subject to the institution’s KVKK procedures.
From stadium to bank vault, courthouse archive to data centre: know who walks in at critical sites, record it, stop it when needed.
Spot the banned person at the turnstile, before they enter.
Everyone entering the file and evidence room is on record.
Closed-circuit access control where no data leaves.
Only the project team enters the sensitive area.
Periodic reports are produced with fully deterministic calculation, and no figure is estimated or rounded. Monthly reports are calculated automatically from the calendar: if today is the 23rd, the report covers days 1-23, and if it is the last day of the month, it covers days 1-30/31.
When a threat is detected, an unauthorised person enters a zone or an unknown person passes through a door, the photo, camera and location of the event arrive on the manager's phone as a push notification with an image. You can see what happened from the lock screen without even opening the app.
📲
Available as PDF in English, Turkish and Arabic. Open it on your phone or in a meeting, and share it freely within your organization. For the user guide, see the Guides page.
Face recognition, field protection, meeting mode, encrypted photo storage, API/Webhook and fully closed-circuit operation: the whole system in 12 pages.
Stadium entrances, bank vault areas, courthouse and treasury archives, high-security facility rooms and more, each explained with illustrations.
The hardware list and installation plan are prepared together, based on your number of cameras and the layout of your site. Fill in the form and our team will get back to you within 1 business day.