ETT Vision AI
AI video analytics for CCTV
AI analytics for CCTV cameras that runs on the customer's own computer, with no cloud: face recognition, Egyptian license plates, safety, swimming pools, child protection and weapon detection, linked to parking, access and attendance.


Screens are in Arabic.
What ETT Vision AI does
51 ready cases in 6 groups
Identity and access, safety and incidents, behavior and panic, swimming pools, child protection, and site and vehicles.
Egyptian license plates
Finds the plate, then reads letters and numbers, confirms by repetition, with allow and deny lists.
Face recognition
Registered people by photo, new person, a deny list, and fake-face detection (a photo or phone in front of the camera).
Safety
Fire and smoke, a person falling, a person still for a long time, intrusion into a zone, crowding, and running or panic.
Swimming pools
People in the water in real time, a person who disappeared inside the pool, suspected drowning, and the pool after hours.
Integration with ETT software
With parking and access, a regular camera reads the plate or face and opens by the same permissions; with Time, staff bus arrival and attendance by face.
Examples by site
At a compound or resort, pool cameras count people in the water and alert on suspected drowning or a child alone near the water, while the gate reads car plates and opens for residents with ETT Access Pro.
At a factory or warehouse: counting cartons and pallets on a line, alerts for fire and smoke or a worker falling, and intrusion into restricted zones. At a car park: free spaces and plate reading with ETT Parking Pro. At a company: face attendance from CCTV with ETT Time Pro.
AI video analytics on your own computer
ETT Vision AI turns existing CCTV cameras into cameras that understand what is happening. The software runs on the customer's own computer, and video, images and faces stay there, with nothing uploaded to an external server.
There is a main screen and monitoring screens on the network, each with a login, permissions and an encrypted connection.
51 ready cases in 6 groups
Each case has a zone, duration and sensitivity set from the screen. The six groups are identity and access, safety and incidents, behaviour and panic, swimming pools, child protection, and site and vehicles. The software includes a case catalog with installation recommendations for each case: megapixels, lens and height.
Egyptian license plate recognition software
The software uses two models: the first finds the plate, and the second reads letters from the right and numbers from the left. A reading is confirmed by repetition, with allow and deny lists.
Face recognition software
Registered people by photo, a new-person alert, a deny list, and fake-face detection when someone holds a photo or a phone in front of the camera.
Drowning detection for swimming pools
A live count of people in the water, an alert when a person disappears inside the pool, suspected drowning with an underwater camera (a person still under the surface), and an alert when someone enters the pool after hours.
Child protection and weapon detection
A child alone near the water, near the exit gate or in a dangerous area, from body shape only and without identifying the child.
Weapon detection: a visible pistol or knife in several consecutive frames raises an alert that security reviews with the image before taking any action.
Safety and counting
Safety: fire and smoke, a person falling, a person still for a long time, intrusion into a zone, crowding, and running or panic.
Counting: people in and out, vehicles and their types, cartons and pallets crossing a line, and free parking spaces on the counters page.
Model accuracy, stated honestly
Accuracy is measured on test images from open datasets, and every case is tested on the site's cameras at handover before anyone relies on it:
- Strong models: Egyptian plate reading, parking spaces (free or occupied), a person falling, visible weapons, and counting cartons and pallets.
- Models tested on site: suspected drowning, vehicle types, swimmers, children, and fire and smoke.
- Experimental models: smoking and fights.
Sensitive alerts support people
Sensitive alerts such as weapons, drowning and children assist the responsible person and do not replace them. Cases without a model do not appear in the software.
Integration with ETT software
With ETT Parking Pro and ETT Access Pro, a regular camera reads the plate or face and opens with the same permission decision. With ETT Time Pro, the staff bus arrival is read from its plate and attendance is taken by face from CCTV. Integration is a licensed option.
Requirements and licensing
- A Windows 10 or 11 (64-bit) computer; processor, memory and graphics card depend on the number of cameras and cases in the site survey.
- Any IP camera with an RTSP stream: Hikvision, Dahua, EZVIZ and others.
- Licensed the ETT way: cases, groups and number of cameras.
How cases are tested at handover
Each case runs on the site's own cameras before anyone relies on it, because lighting, camera angle and height affect the result. The software includes installation recommendations for each case, so any camera that needs adjusting shows up during the survey.
Privacy in detail
Video, images and faces stay on the customer's computer, with nothing uploaded to an external server. Monitoring screens on the network sign in with a user name, permissions and an encrypted connection. For child protection the software works on body shape only, without identifying the child.
Events and registered people
Every event is recorded with the image, time and place, so security can go back to any alert and see what happened. Registered people are added by photo, including a deny list that raises an alert as soon as one of them appears.
Who uses video analytics
- Compounds and resorts: swimming pools, child protection and plate reading at the gate.
- Factories and warehouses: safety, counting and intrusion.
- Car parks: free spaces and plate reading.
- Schools: child protection and safety.
- Companies: face attendance and staff bus arrival.
Why ETT Vision AI
- It works with your existing cameras if they stream RTSP.
- Privacy: all video and faces stay on your computer.
- Egyptian plate reading with letters and numbers.
- 51 ready cases with zone, duration and sensitivity set from the screen.
- Integration with parking, access and attendance, so a camera can open a gate with the same permissions.
- Honest accuracy: every case is tested on your cameras before you rely on it.
How to order ETT Vision AI
- Send us the number of cameras and their brands.
- Tell us the cases you need: plates, faces, pools or safety.
- We carry out a site survey to set the computer specifications and camera positions, then send a quotation.
Inside the software


Supported devices
| Device | Role with the software |
|---|---|
| Any IP camera with an RTSP stream | Hikvision, Dahua, EZVIZ and others; the megapixels, lens and height for each case are listed in the software |
| Windows 10 or 11 computer | Specifications are set by the number of cameras and cases during the site survey |
Terms you will see in the software
- RTSPThe way an IP camera streams video; most modern cameras support it.
- Fake-face detectionThe software spots someone trying to pass with a photo or a phone in front of the camera.
- No cloudAll analysis runs on the customer's computer, and no video leaves the site.
Who it is for
- Compounds and resorts with pools
- Factories and warehouses
- Car parks
- Schools
Frequently asked questions
What does ETT Vision AI do?
It analyzes CCTV cameras with AI and raises alerts for cases such as a denied person, suspected drowning, fire or a weapon, and reads Egyptian license plates.
Is the video uploaded to the internet?
No. Video, images and faces stay on the customer's computer; nothing is uploaded to an external server.
Which cameras does it work with?
Any IP camera with an RTSP stream, such as Hikvision, Dahua and EZVIZ.
How much does ETT Vision AI cost?
Pricing follows the license: cases, groups and number of cameras, with computer specifications set in the site survey.
Do I need new cameras?
Usually not. Any IP camera with an RTSP stream works, and the software lists the right megapixels, lens and height for each case.
Does it read Egyptian plates?
Yes. It finds the plate and reads the letters and numbers, confirming the reading by repetition.
Is fire detection reliable?
The fire and smoke model in this version is one of the models tested on site before it is relied on.
Does it identify children?
No. Child protection works from body shape only, without identifying the child.
Does a weapon alert come from one image?
No. The pistol or knife must appear in several consecutive frames, and security reviews the alert with the image before acting.
Can I start with a few cases and add more?
Yes, the license is by cases, groups and cameras, so you can start with what you need.