Developer Platform

Build Video Analytics Into Anything

CVEDIA-RT is a modular AI inference engine with REST API, native SDKs, and a plugin architecture that runs on 10+ hardware accelerators. Embed it in cameras, integrate with your platform, or build custom solutions.

View Documentation GitHub SDK

REST API

# Create an analytics instance
curl -X POST "http://localhost:3546/v1/core/instance" \
    -H "Content-Type: application/json" \
    -d '{"name": "cam-1", "solution": "securt"}'

# Set RTSP input source
curl -X POST "http://localhost:3546/v1/core/instance/cam-1/input" \
  -H "Content-Type: application/json" \
  -d '{"type": "RTSP", "uri": "rtsp://192.168.1.100:554/stream"}'

# Start the instance
curl -X POST "http://localhost:3546/v1/core/instance/cam-1/start"

# Stream events
curl "http://localhost:3546/v1/core/instance/cam-1/consume_events"

The Stack

Everything You Need to Integrate

CVEDIA-RT is the same engine that powers our VMS plugins—now available for your own products and platforms.

RT Server

Headless service with REST API and Web Panel. Run as a service on Windows or Linux, manage instances remotely, consume events via API.

REST API Web Admin Multi-instance Docker

Native SDK

Language bindings for C++, C, C#, and Python. Interact directly with the inference engine, extend with custom plugins, embed in your application.

C++ API Python C#/.NET Headers

REST API

Three API namespaces: Core (instance management), SecuRT (security analytics), and ONVIF (camera discovery). Interactive docs included.

Core API SecuRT API ONVIF API SSE Events

Model Forge

One model definition, compiled for any target hardware. TensorRT, OpenVINO, Hailo, RKNN—Model Forge handles optimization automatically.

Auto-optimization Multi-target INT8/FP16 Batching

Integration Options

Choose How You Build

From quick HTTP integrations to deep embedded development—pick the approach that fits your architecture.

REST API

HTTP-based integration for any language or platform. Manage instances, push frames, consume events.

Any HTTP client

Native SDK

Direct engine access for maximum performance. Build custom inference plugins or embed in applications.

C++ C C# Python

Plugin Architecture

Extend CVEDIA-RT with custom inference engines, input sources, output handlers, and processing modules.

C++ plugins Lua scripts

10+ AI Accelerators, One Codebase

Write once, deploy anywhere. CVEDIA-RT's plugin architecture abstracts hardware complexity so your integration works across platforms.

TensorRT

NVIDIA GPU

OpenVINO

Intel CPU/GPU/VPU

Hailo-8/15

Edge AI

RKNN

Rockchip NPU

SNPE

Qualcomm

CVFlow

Ambarella

Blaize GSP

Ultra-low power

DeepX

Edge inference

Plus ARM NN, Jetson (Orin/Xavier/Nano), SigmaStar, and more. View full platform list →

Use Cases

What People Build With CVEDIA

From camera manufacturers to enterprise platforms—the same engine powering diverse products across industries.

OEM Camera Analytics

Embed analytics directly into camera firmware. Run on-device inference with optimized models for your specific hardware.

Custom VMS Plugins

Build your own VMS integration using the same architecture as our Milestone and Nx plugins.

Edge Appliances

Pre-configure CVEDIA-RT on your hardware. Sell turnkey analytics boxes with your branding.

Cloud Analytics

Deploy RT Server on AWS, Azure, or GCP. Scale instances dynamically with your workload.

Traffic & ITS Systems

Integrate vehicle detection, classification, and counting into existing traffic management infrastructure.

Defense & Aerospace

Custom models for rare targets, optimized for specific sensors and operational requirements.

Event Consumer

const events = await fetch('http://localhost:3546/v1/core/instance/cam-1/consume_events').then(r => r.json());

for (const event of events) {
  const eventData = JSON.parse(event.jsonObject);

  switch(event.dataType) {
    case 'event-intrusion-v1':
      handleIntrusion(eventData);
      break;
    case 'event-line-crossing-v1':
      handleLineCrossing(eventData);
      break;
    case 'event-area-enter-v1':
      handleAreaEnter(eventData);
      break;
    case 'event-area-exit-v1':
      handleAreaExit(eventData);
      break;
  }
}

API Reference

Three Namespaces, Full Control

Interactive documentation at localhost:3546 when RT Server is running. Test endpoints directly from the docs.

Core API

Instance lifecycle, input/output control, frame pushing, event consumption, system operations.

SecuRT API

Security analytics configuration—zones, tripwires, detection rules, sensitivity tuning.

ONVIF API

Camera discovery, device management, stream configuration via ONVIF protocol.

Getting Started

From Zero to Detection in Minutes

Install RT Server

Download for Windows or Linux. Docker images available for containerized deployments.

Create an Instance

Use the REST API or Web Panel to configure your first analytics instance with a video source.

Consume Events

Subscribe to detection events via REST API, SSE, MQTT, or webhook. Integrate with your system.

Get CVEDIA-RT

Free evaluation license included. Production licensing available for commercial deployments.

Get CVEDIA-RT

Need Help Getting Started?

Our engineering team can help with integration architecture, custom model requirements, and production deployment planning.

Read the Docs Contact Engineering