PatrickAttankurugu/ghana-card-detector
Domain:
digital infrastructureRecord type:
softwaremodel
Creator:
Pat
Host:
NPM package
# Ghana Card Detector
A JavaScript/TypeScript package for detecting Ghana cards in images and video streams using TensorFlow.js and a custom-trained YOLOv8 model.
## Features
- Real-time Ghana card detection
- Support for both image and video input
- High accuracy (97.48% precision, 100% recall)
- TypeScript support
- Built-in visualization tools
- Memory-efficient model loading
- Customizable detection parameters
## Installation
```bash
npm install ghana-card-detector
Basic Usage
Typescript
import { GhanaCardDetector } from 'ghana-card-detector';
// Initialize detector
const detector = new GhanaCardDetector();
await detector.initialize();
// Detect from image
const img = document.querySelector('img');
const results = await detector.detect(img);
// Process results
results.forEach(detection => {
console.log('Detection:', {
confidence: detection.confidence,
boundingBox: detection.box
});
});
Example with Video Stream
typescriptCopyimport { GhanaCardDetector } from 'ghana-card-detector';
async function setupDetector() {
// Initialize detector
const detector = new GhanaCardDetector();
await detector.initialize();
// Access camera
const stream = await navigator.mediaDevices.getUserMedia({ video: true });
const video = document.querySelector('video');
video.srcObject = stream;
// Create canvas for visualization
const canvas = document.createElement('canvas');
document.body.appendChild(canvas);
// Detection loop
async function detectFrame() {
const detections = await detector.detect(video);
// Draw results
detector.drawDetections(canvas, detections);
requestAnimationFrame(detectFrame);
}
detectFrame();
}
Advanced Configuration
typescriptCopyconst detector = new GhanaCardDetector({
modelUrl: 'custom-model-url', // Custom model URL
version: '1.0.0', // Model version
scoreThreshold: 0.5, // Detection confidence threshold
maxDetections: 5 // Maximum numb …