UNPKG

react-native-vision-camera-face-detector

Version:

Frame Processor Plugin to detect faces using MLKit Vision Face Detector for React Native Vision Camera!

188 lines (169 loc) 5.87 kB
"use strict"; Object.defineProperty(exports, "__esModule", { value: true }); exports.Camera = Camera; var _react = _interopRequireWildcard(require("react")); var _reactNativeVisionCamera = require("react-native-vision-camera"); var _reactNativeWorkletsCore = require("react-native-worklets-core"); var _FaceDetector = require("./FaceDetector"); var _jsxRuntime = require("react/jsx-runtime"); function _interopRequireWildcard(e, t) { if ("function" == typeof WeakMap) var r = new WeakMap(), n = new WeakMap(); return (_interopRequireWildcard = function (e, t) { if (!t && e && e.__esModule) return e; var o, i, f = { __proto__: null, default: e }; if (null === e || "object" != typeof e && "function" != typeof e) return f; if (o = t ? n : r) { if (o.has(e)) return o.get(e); o.set(e, f); } for (const t in e) "default" !== t && {}.hasOwnProperty.call(e, t) && ((i = (o = Object.defineProperty) && Object.getOwnPropertyDescriptor(e, t)) && (i.get || i.set) ? o(f, t, i) : f[t] = e[t]); return f; })(e, t); } // types /** * Create a Worklet function that persists between re-renders. * The returned function can be called from both a Worklet context and the JS context, but will execute on a Worklet context. * * @param {function} func The Worklet. Must be marked with the `'worklet'` directive. * @param {DependencyList} dependencyList The React dependencies of this Worklet. * @returns {UseWorkletType} A memoized Worklet */ function useWorklet(func, dependencyList) { const worklet = _react.default.useMemo(() => { const context = _reactNativeWorkletsCore.Worklets.defaultContext; return context.createRunAsync(func); }, dependencyList); return worklet; } /** * Create a Worklet function that runs the giver function on JS context. * The returned function can be called from a Worklet to hop back to the JS thread. * * @param {function} func The Worklet. Must be marked with the `'worklet'` directive. * @param {DependencyList} dependencyList The React dependencies of this Worklet. * @returns {UseRunInJSType} a memoized Worklet */ function useRunInJS(func, dependencyList) { return _react.default.useMemo(() => _reactNativeWorkletsCore.Worklets.createRunOnJS(func), dependencyList); } /** * Vision camera wrapper * * @param {ComponentType} props Camera + face detection props * @returns */ function Camera({ ref, faceDetectionOptions, faceDetectionCallback, skiaActions, ...props }) { /** * Is there an async task already running? */ const isAsyncContextBusy = (0, _reactNativeWorkletsCore.useSharedValue)(false); const faces = (0, _reactNativeWorkletsCore.useSharedValue)('[]'); const { detectFaces, stopListeners } = (0, _FaceDetector.useFaceDetector)(faceDetectionOptions); (0, _react.useEffect)(() => { return () => stopListeners(); }, []); /** * Throws logs/errors back on js thread */ const logOnJs = _reactNativeWorkletsCore.Worklets.createRunOnJS((log, error) => { if (error) { console.error(log, error.message ?? JSON.stringify(error)); } else { console.log(log); } }); /** * Runs on detection callback on js thread */ const runOnJs = useRunInJS(faceDetectionCallback, [faceDetectionCallback]); /** * Async context that will handle face detection */ const runOnAsyncContext = useWorklet(frame => { 'worklet'; try { faces.value = JSON.stringify(detectFaces(frame)); // increment frame count so we can use frame on // js side without frame processor getting stuck frame.incrementRefCount(); runOnJs(JSON.parse(faces.value), frame).finally(() => { 'worklet'; // finally decrement frame count so it can be dropped frame.decrementRefCount(); }); } catch (error) { logOnJs('Execution error:', error); } finally { frame.decrementRefCount(); isAsyncContextBusy.value = false; } }, [detectFaces, runOnJs]); /** * Detect faces on frame on an async context without blocking camera preview * * @param {Frame} frame Current frame */ function runAsync(frame) { 'worklet'; if (isAsyncContextBusy.value) return; // set async context as busy isAsyncContextBusy.value = true; // cast to internal frame and increment ref count const internal = frame; internal.incrementRefCount(); // detect faces in async context runOnAsyncContext(internal); } /** * Skia frame processor */ const skiaFrameProcessor = (0, _reactNativeVisionCamera.useSkiaFrameProcessor)(frame => { 'worklet'; frame.render(); skiaActions(JSON.parse(faces.value), frame); runAsync(frame); }, [runOnAsyncContext, skiaActions]); /** * Default frame processor */ const cameraFrameProcessor = (0, _reactNativeVisionCamera.useFrameProcessor)(frame => { 'worklet'; runAsync(frame); }, [runOnAsyncContext]); /** * Camera frame processor */ const frameProcessor = (() => { const { autoMode } = faceDetectionOptions ?? {}; if (!autoMode && !!skiaActions) return skiaFrameProcessor; return cameraFrameProcessor; })(); // // use bellow when vision-camera's // context creation issue is solved // // /** // * Runs on detection callback on js thread // */ // const runOnJs = useRunOnJS( faceDetectionCallback, [ // faceDetectionCallback // ] ) // const cameraFrameProcessor = useFrameProcessor( ( frame ) => { // 'worklet' // runAsync( frame, () => { // 'worklet' // runOnJs( // detectFaces( frame ), // frame // ) // } ) // }, [ runOnJs ] ) return /*#__PURE__*/(0, _jsxRuntime.jsx)(_reactNativeVisionCamera.Camera, { ...props, ref: ref, frameProcessor: frameProcessor, pixelFormat: "yuv" }); } //# sourceMappingURL=Camera.js.map