Detect objects in images
Identify the main objects in an image
Identify labels in an image with a score threshold
Identify objects in images
Detect marine vessels in images
Upload images to detect objects
Find objects in images using text descriptions
Detect objects in images and return coordinates
Identify and label objects in images
Detect gestures in images and video
Detect and measure areas of objects in images
Identify objects in images using text queries
Find and highlight characters in images
Transformers.js is a JavaScript library designed for object detection tasks. Built on top of TensorFlow.js, it provides an efficient way to detect objects within images. The library leverages transformer-based architectures to deliver high accuracy and flexibility for various object detection use cases.
npm install transformers.js
import { TransformerDetector } from 'transformers.js';
const detector = new TransformerDetector('model_name');
const image = document.getElementById('image');
const predictions = await detector.detect(image);
predictions.forEach(prediction => {
console.log(`Detected ${prediction.label} at ${prediction.bbox}`);
});
What models are supported by Transformers.js?
Transformers.js supports popular object detection models like DETR, Deformable DETR, and YOLOS.
Can Transformers.js run on mobile browsers?
Yes, Transformers.js is optimized for mobile browsers and supports TensorFlow.js's mobile-friendly backend.
How accurate is Transformers.js compared to other libraries?
Transformers.js leverages state-of-the-art transformer architectures, often outperforming traditional CNN-based detectors in accuracy.