Arya Vision

Detailed overview

Arya Vision is AryaNet’s video analytics platform for sites that want local control of imagery. It connects IP cameras over RTSP, runs detection and tracking on GPU on a Windows server, and provides a web UI plus API for monitoring, alerts, and evidence — without a mandatory video cloud.

In the field, Arya Vision connects to existing camera infrastructure and turns video into an operational stream: detected objects, tracked paths, stored events. It is built for industrial sites, perimeters, entrances, and campuses where operators need fast reaction and clear evidence — not only passive NVR recording.

A typical architecture separates the analysis pipeline (continuous processing on active cameras) from the web layer (UI and API). Everything runs natively on Windows with GPU hardware so data and decisions stay on your network. Arya Vision is software — it does not replace fire detection (IDSA) design or installation; it complements physical security with video intelligence.

What it solves

IP cameras generate more video than a small team can watch 24/7. Arya Vision reduces noise: it extracts relevant events (presence, movement of interest, on-frame tracking) and surfaces them in the UI.

For operators, that means less time on empty screens and more evidence when an incident happens. For management, it means a measurable system: which cameras are active, which pipeline is running, which alerts were raised.

Architecture (brief)

Cameras send RTSP streams to the analysis pipeline. The pipeline runs detection and tracking on GPU, writes events/alerts to the database, and can keep frames or clips for evidence.

The web layer owns the monitoring UI and API: health, visualization, operational admin. The main production path is native Windows — not a container-first stack.

Input: IP / RTSP cameras (including Hikvision ecosystem).
Processing: GPU detection + tracking (dedicated pipeline).
Data: events and metadata in a relational store; files for frames/clips.
Output: web UI + API for operations and integrations.

Features by category

Cameras and ingest

IP camera integration over RTSP on configurable channels.
Hikvision ecosystem support (ISAPI / ONVIF in admin tools).
Operational preference for substreams when you want lower GPU load.
Camera config in database / config files — not hard-coded in the UI.

Detection and tracking

Continuous pipeline on cameras marked active.
GPU detection with modern detector-class models (e.g. RT-DETR line).
Object tracking (e.g. ByteTrack) for continuity across frames.
Room for parallel detection lines (e.g. OpenVINO / YOLO roles in lab or other duties).

Operations and evidence

Web UI for monitoring and pipeline health.
API for health checks and integrations.
Alerts with frames / clips for later review.
Windows services for web and pipeline — controlled restart after updates.

Capabilities summary

IP camera integration over RTSP (including Hikvision).
GPU detection and tracking on active streams.
Continuous pipeline on configured cameras.
Web UI + API for status and events.
Alerts with frames / clips for evidence.
Runs on an on‑premise Windows server with GPU.
Camera admin via dedicated tools (test, streams, diagnose).
Separate from IDSA installation services — software product.

How it works — steps

Inventory cameras and RTSP channels useful for AI (usually substream).
Install / configure Arya Vision on the Windows GPU server.
Enable the pipeline on selected cameras and verify service health.
Operators use the UI for monitoring; alerts accumulate as evidence.
Iterate: add cameras, tune load, connect the API to other systems if needed.

Use cases

Perimeter and access gates on an industrial site.
Corridors, entrances, and critical zones on a campus / office.
Assisted surveillance where an NVR alone is not enough for reaction.
Post-incident evidence with frames and metadata — not only “rewind the tape”.

Integrations and stack

Existing IP / RTSP cameras.
Hikvision (ISAPI / ONVIF via admin tools).
NVIDIA GPU + ONNX Runtime / CUDA on Windows.
MSSQL for configuration and events.
HTTP API for monitoring and later integrations.

Who it’s for · What you get · What it is not

Who it’s for

Physical security teams, site operators, campuses, industry, and organizations that want video analytics on their own infrastructure, with local data.
What you get

Tracking pipeline, Windows services, web UI, API, local storage for alerts/frames, and operations documentation — on‑premise delivery.
What it is not

Arya Vision is not a classic NVR and does not replace IDSA design/installation. It is not a mandatory cloud video SaaS. The public sheet does not include lab details (internal IPs, passwords, server paths).