luicfrr / react-native-vision-camera-face-detector

Vision Camera Frame Processor Plugin to detect faces using MLKit Face Detector
https://www.npmjs.com/package/react-native-vision-camera-face-detector
MIT License
131 stars 22 forks source link
face-detection face-detector vision-camera

πŸ“š Introduction

react-native-vision-camera-face-detector is a React Native library that integrates with the Vision Camera module to provide face detection functionality. It allows you to easily detect faces in real-time using device's front and back camera.

If you like this package please give it a ⭐ on GitHub.

πŸ—οΈ Features

🧰 Installation

yarn add react-native-vision-camera-face-detector

Then you need to add react-native-worklets-core plugin to babel.config.js. More details here.

πŸ’‘ Usage

Recommended way:

import { 
  StyleSheet, 
  Text, 
  View 
} from 'react-native'
import { 
  useEffect, 
  useState,
  useRef
} from 'react'
import {
  Frame,
  useCameraDevice
} from 'react-native-vision-camera'
import {
  Face,
  Camera,
  FaceDetectionOptions
} from 'react-native-vision-camera-face-detector'

export default function App() {
  const faceDetectionOptions = useRef<FaceDetectionOptions>( {
    // detection options
  } ).current

  const device = useCameraDevice('front')

  useEffect(() => {
    (async () => {
      const status = await Camera.requestCameraPermission()
      console.log({ status })
    })()
  }, [device])

  function handleFacesDetection(
    faces: Face[],
    frame: Frame
  ) { 
    console.log(
      'faces', faces.length,
      'frame', frame.toString()
    )
  }

  return (
    <View style={{ flex: 1 }}>
      {!!device? <Camera
        style={StyleSheet.absoluteFill}
        device={device}
        faceDetectionCallback={ handleFacesDetection }
        faceDetectionOptions={ faceDetectionOptions }
      /> : <Text>
        No Device
      </Text>}
    </View>
  )
}

Or use it following vision-camera docs:

import { 
  StyleSheet, 
  Text, 
  View 
} from 'react-native'
import { 
  useEffect, 
  useState,
  useRef
} from 'react'
import {
  Camera,
  useCameraDevice,
  useFrameProcessor
} from 'react-native-vision-camera'
import { 
  Face,
  runAsync,
  useFaceDetector,
  FaceDetectionOptions
} from 'react-native-vision-camera-face-detector'
import { Worklets } from 'react-native-worklets-core'

export default function App() {
  const faceDetectionOptions = useRef<FaceDetectionOptions>( {
    // detection options
  } ).current

  const device = useCameraDevice('front')
  const { detectFaces } = useFaceDetector( faceDetectionOptions )

  useEffect(() => {
    (async () => {
      const status = await Camera.requestCameraPermission()
      console.log({ status })
    })()
  }, [device])

  const handleDetectedFaces = Worklets.createRunOnJS( (
    faces: Face[]
  ) => { 
    console.log( 'faces detected', faces )
  })

  const frameProcessor = useFrameProcessor((frame) => {
    'worklet'
    runAsync(frame, () => {
      'worklet'
      const faces = detectFaces(frame)
      // ... chain some asynchronous frame processor
      // ... do something asynchronously with frame
      handleDetectedFaces(faces)
    })
    // ... chain frame processors
    // ... do something with frame
  }, [handleDetectedFaces])

  return (
    <View style={{ flex: 1 }}>
      {!!device? <Camera
        style={StyleSheet.absoluteFill}
        device={device}
        isActive={true}
        frameProcessor={frameProcessor}
      /> : <Text>
        No Device
      </Text>}
    </View>
  )
}

As face detection is a heavy process you should run it in an asynchronous thread so it can be finished without blocking your camera preview. You should read vision-camera docs about this feature.

Face Detection Options

Option Description Default
performanceMode Favor speed or accuracy when detecting faces. fast
landmarkMode Whether to attempt to identify facial landmarks: eyes, ears, nose, cheeks, mouth, and so on. none
contourMode Whether to detect the contours of facial features. Contours are detected for only the most prominent face in an image. none
classificationMode Whether or not to classify faces into categories such as 'smiling', and 'eyes open'. none
minFaceSize Sets the smallest desired face size, expressed as the ratio of the width of the head to width of the image. 0.15
trackingEnabled Whether or not to assign faces an ID, which can be used to track faces across images. Note that when contour detection is enabled, only one face is detected, so face tracking doesn't produce useful results. For this reason, and to improve detection speed, don't enable both contour detection and face tracking. false
autoScale Should auto scale face bounds, contour and landmarks on native side? If this option is disabled all detection results will be relative to frame coordinates, not to screen/preview. You shouldn't use this option if you want to draw on screen using Skia Frame Processor. See this and this for more details. false
windowWidth * Required if you want to use autoScale. You must handle your own logic to get screen sizes, with or without statusbar size, etc... 1.0
windowHeight * Required if you want to use autoScale. You must handle your own logic to get screen sizes, with or without statusbar size, etc... 1.0

πŸ”§ Troubleshooting

Here is a common issue when trying to use this package and how you can try to fix it:

If you find other errors while using this package you're wellcome to open a new issue or create a PR with the fix.

πŸ‘· Built With

πŸ”Ž About

This package was tested using the following:

Min O.S version:

Make sure to follow tested versions and your device is using the minimum O.S version before opening issues.

πŸ“š Author

Made with ❀️ by luicfrr