# YOLO v5 in Node.JS

Object Detection using YOLO for Node.js backend.  

***ES6 compatibile***


## Installation  

`npm i yolov5`

It also installs the required dependencies: `tfjs`, `tfjs-node` and `regenerator-runtime`  

For more info on tensorflow backend in node.js:  
(https://github.com/tensorflow/tfjs/tree/master/tfjs-node)
(https://github.com/tensorflow/tfjs)


## Usage  

```javascript

require('regenerator-runtime');
const yolov5 =  require('yolov5');
// import {YOLOv5s} from 'yolov5';


const main = async() => {	
	let predictions;

	const yolo = yolov5;		// Create an instance of the model 
	await yolo.load();			// Load Model 

	if (yolo.model != '') {
		console.log("Model loaded \n");
		const result = await yolo.predict(image);		// Input image must be in 640x640 format 
		predictions = yolo.getDetections(result);		// Get bounding box(s) and class data 
	} else {
		return console.error("Model not found \n");
	}

	if (predictions.length > 0) console.log("Predictions:", predictions);
	
	yolo.dispose;
};

main();

```

View model parameters, methods, functions: `console.log(yolo.details())`  


### Output  

Prediction format: 
```javascript
{
	"bbox": [409, 1, 639, 169],		// [x0, y0, x1, y1]
	"score": 0.6489,				// probability score
	"class": "car"					// Object class name
}
```

**Notes**  

* Current backend support only for CPU via `tfjs-node`.  

* Insert `regenerator-runtime` or its equivalent for async-await based functional implementation.