YazSis VisionAIA YazSis brandYazSis VisionAI is a smart camera management platform designed for IP cameras in factories, warehouses and logistics sites. Scenarios such as people counting, area occupancy, zone violation and helmet/vest compliance are defined without writing code; events flow to the dashboard, to notifications and to scheduled reports. By design, footage is processed on the on-site server and only event data goes to the central system. It does not perform face recognition or identification.
In developmentThe management dashboard, API and edge agent are ready in an end-to-end working prototype (v0.1) and run on simulated data today. The live camera feed and the detection model will come into play with the first pilot facility.
Rules decide who can see what and what needs approval; a scenario does not run until it is approved.
The facility, building and area tree is set up; cameras are added one by one or via CSV. Camera passwords are never shown in the dashboard and are stored encrypted.
Restricted zones, counting lines, masks and seating areas are drawn on the camera frame.
A scenario is built from templates and parameters, without writing code; it does not run until the data controller representative approves its purpose.
The edge agent runs on site; it sends event and metric data, not footage, to the center, and buffers locally if the connection drops. Analysis with real cameras starts with the pilot.
Uncertain events go to the review queue; reports are emailed, limited to the areas each recipient is authorized for.
The platform’s rules answer up front the five questions asked most often about camera analytics.
Face recognition, identification and cross-camera person tracking are deliberately out of scope; the scenario wizard answers such a request with a “not allowed” warning.
Every new scenario starts out pending approval; a scenario whose scope widens goes back for approval, and decisions are written to the audit log.
By design, analysis runs on the on-site server; the edge agent sends only event and metric data to the center.
Uncertain events go to a person for confirmation; alarms alone should not be used as grounds for disciplinary action.
Minimum duration, repeat cooldown, quiet hours, digest mode and escalation cut down unnecessary notifications.
The badges are honest: “Working in prototype” means the feature works in the dashboard with simulated data, “Opens with pilot” means it comes into play with real cameras and installation, and “Roadmap” means it has not been developed yet.
Facility, building and area tree; adding cameras, bulk import via CSV, search and filtering. Camera passwords are never returned in responses and are stored encrypted.
Templates for zone violation, line counting, occupancy, PPE, dwell time and camera health; a six-step wizard.
A new scenario waits for approval; only the data controller representative or an administrator can approve it, and it goes back for approval if its scope widens.
Live event feed, filtering, bulk actions, true and false alarm marking; uncertain events go to a person for confirmation.
Rules by event type, area, severity and time; repeat cooldown, quiet hours, digest mode and escalation to a higher group when an event goes unreviewed.
Six report types, daily, weekly and monthly; HTML and Excel output is sent by email, and each recipient sees only the areas they are authorized for.
Role- and area-based permissions, an audit log of every write action, adjustable retention periods and deletion requests.
The on-site agent pulls its configuration from the center, sends event and metric data in batches and buffers locally during an outage. Today it runs in simulation mode.
Rules for line crossing, occupancy, zone violation, dwell time and camera health are ready and tested; the first measurement with real cameras and a detection model takes place at the pilot site.
The scenario and reporting side is ready; the required detection model will be prepared for the conditions of the pilot site. Accuracy is not shared until it has been measured on site.
Automatic blurring of faces in event images before they are stored or shared.
Defining a new situation with example images; the flow for collecting examples, labeling and KVKK approval exists in the prototype, but model training does not yet.
The frames are taken from the working prototype; the flows work, the numbers and camera frames are simulated.
Active cameras, today’s events, pending reviews, occupancy and scenario status on one screen; the numbers are simulated data.
It is a smart camera management platform designed for IP cameras in factories, warehouses and logistics sites. It aims to let you define scenarios such as people counting, area occupancy, zone violation and helmet/vest compliance without writing code, and to follow events on the dashboard, in notifications and in scheduled reports. It is a YazSis brand.
The management dashboard, API and edge agent form an end-to-end working prototype (v0.1) that runs on simulated data. The live camera feed and the detection model will come into play with the first pilot facility; accuracy and speed figures will be measured and shared then.
No. Face recognition, identification and cross-camera person tracking are deliberately out of scope. Automatic face blurring in images is on the roadmap; today the edge agent does not send footage to the center.
By design, analysis runs on the on-site server and the edge agent sends only event and metric data to the center. Sending footage to cloud AI is off by default; turning it on requires explicit approval and is recorded in the audit log.
The goal is to work with existing IP cameras that provide an RTSP stream. Which cameras are suitable, along with viewing angle and resolution, is checked camera by camera in discovery before the pilot. Recordings stay on your existing network video recorder (NVR); VisionAI does not replace it.
Write to us through the contact form and choose the topic “YazSis VisionAI”. We review your facility, camera infrastructure and priority scenarios together, then talk through the pilot scope.
YazSis CRM AISales, support and licensing in one hubKurumsalPortal33 separate modules / programs, one platform
Super Randevu24/7 AI appointment assistant
SahaSesi AISite logs and evening report from WhatsApp
YazSis Akıllı İhaleManage tenders and bids in a single recordDÖNGÜDigital product passport operating systemKarbonİzCarbon footprint management platform