Smart Traffic Analyzer SDK

A client-side JavaScript SDK for traffic analysis using ONNX models.
Supports vehicle detection, road segmentation, lane segmentation, congestion analysis, and traffic violations directly in the browser.

Features

• Detection – Detect vehicles and pedestrians in images or video.
• Segmentation – Road/lane segmentation for context.
• Vehicle Count – Count vehicles per class.
• Congestion Analysis – Estimate traffic density.
• Violation Detection – Detect lane or red light violations.
• SDK Status – Check if models are loaded.

Installation

Using npm:

npm install smart-traffic-analyzer

Using CDN

<script type="module">
  import { TrafficSDK } from 'https://cdn.jsdelivr.net/gh/username/smart-traffic-analyzer/dist/sdk.js';
</script>

Quick Start

Vanilla JavaScript

<script type="module">
    import { TrafficSDK } from './dist/sdk.js';

    const sdk = new TrafficSDK();

    (async () => {
    await sdk.init();

    // Detect vehicles in a video element
    const video = document.getElementById('traffic-video');
    const result = await sdk.detect(video);
    console.log(result);

    // Road segmentation (run once)
    const segmentation = await sdk.segment(video);
    console.log(segmentation);

    // Vehicle count
    const counts = await sdk.vehicleCount(video);
    console.log(counts);

    // Congestion
    console.log(sdk.getCongestionLevel());

    // Traffic violations
    console.log(sdk.detectViolations());

    // SDK status
    console.log(sdk.status());
    })();
</script>

React Example

import { useEffect } from 'react';
import { TrafficSDK } from 'smart-traffic-analyzer';

export default function App() {
  useEffect(() => {
    const sdk = new TrafficSDK();

    const runAnalysis = async () => {
      await sdk.init();
      const video = document.getElementById('traffic-video');
      const detection = await sdk.detect(video);
      console.log(detection);
    };

    runAnalysis();
  }, []);

  return <video id="traffic-video" autoPlay />;
}

Notes

• The SDK loads ONNX models from CDN; make sure you have a stable internet connection.
• Segmentation only needs to run once per scene to reduce CPU/GPU usage.
• Works in modern browsers with WebGL/WebAssembly support.

License

MIT © Amoke Emmanuel Chinonye