@avatune/facial-hair-predictor

Source: packages/facial-hair-predictor/README.md

Browser-based facial hair prediction using TensorFlow.js. Classifies faces into 2 categories: none (clean-shaven) or facial_hair (beard, mustache, etc.).

Tiny model (~2.3MB) with fast loading and inference in the browser.

Installation

npm

npm install @avatune/facial-hair-predictor @tensorflow/tfjs

yarn

yarn add @avatune/facial-hair-predictor @tensorflow/tfjs

pnpm

pnpm add @avatune/facial-hair-predictor @tensorflow/tfjs

bun

bun add @avatune/facial-hair-predictor @tensorflow/tfjs

Usage

import { createFacialHairPredictor } from '@avatune/facial-hair-predictor'

// Uses jsDelivr CDN by default - no setup required!
const predictor = createFacialHairPredictor()
await predictor.loadModel()

const result = await predictor.predictFromImage(imageElement)
console.log(result)
// {
//   facialHair: 'facial_hair',
//   confidence: 0.92,
//   probabilities: { none: 0.08, facial_hair: 0.92 },
//   faceDetected: true
// }

Model Files

By default, models are loaded from jsDelivr CDN (https://cdn.jsdelivr.net/npm/@avatune/facial-hair-predictor@1/dist/model). No setup required!

Self-hosting (Optional)

If you prefer to self-host the model files, copy them from dist/model/ to your public directory:

• model.json - Model architecture and weights manifest
• classes.json - Class labels
• group1-shard1of1.bin - Model weights

Then pass the path to the predictor:

const predictor = createFacialHairPredictor('/models/facial-hair')

Setup with Vite (Optional)

import { copyFileSync, mkdirSync, readdirSync } from 'node:fs'
import { join } from 'node:path'
import { defineConfig } from 'vite'

export default defineConfig({
  plugins: [
    {
      name: 'copy-tfjs-models',
      buildStart() {
        const srcDir = join(__dirname, 'node_modules', '@avatune', 'facial-hair-predictor', 'dist', 'model')
        const destDir = join(__dirname, 'public', 'models', 'facial-hair')

        mkdirSync(destDir, { recursive: true })

        copyFileSync(join(srcDir, 'model.json'), join(destDir, 'model.json'))
        copyFileSync(join(srcDir, 'classes.json'), join(destDir, 'classes.json'))

        const files = readdirSync(srcDir)
        for (const file of files) {
          if (file.endsWith('.bin')) {
            copyFileSync(join(srcDir, file), join(destDir, file))
          }
        }

        console.log('✓ Copied facial-hair model to public/models')
      },
    },
  ],
})

API

Structure

createFacialHairPredictor(modelDir?: string)

Parameters:

• modelDir (optional) - Path to directory containing model files. Defaults to jsDelivr CDN

Functions

loadModel(): Promise<void>

Loads the TFJS model and class labels. Call this once before making predictions.

predict(imageTensor: tf.Tensor3D): Promise<FacialHairResult>

Predicts facial hair presence from an image tensor.

Parameters:

• imageTensor - Normalized RGB image tensor [H, W, 3] with values in range [0, 1]

Returns:

{
  facialHair: string     // Predicted class: 'none' | 'facial_hair'
  confidence: number     // Confidence score [0, 1]
  probabilities: Record<string, number>  // Scores for all classes
}

predictFromImage(image): Promise<FacialHairResult>

Predicts facial hair from an image element. Automatically detects and crops the lower face area for better accuracy.

Parameters:

• image - HTMLImageElement, HTMLVideoElement, or HTMLCanvasElement

Returns: Same as predict(), plus faceDetected: boolean

Model Details

• Architecture: MobileNetV2-based CNN
• Input: 128x128 RGB images
• Classes: 2 (none, facial_hair)
• Training: CelebA dataset
• Accuracy: ~83%
• Format: TensorFlow.js with uint8 quantization

License

See LICENSE.md for license information.