Android

关注公众号 jb51net

关闭
首页 > 软件编程 > Android > Android人脸识别

Android基于虹软(ArcSoft)实现人脸识别

作者:Great_7

人工智能时代快速来临,其中人脸识别是当前比较热门的技术,在国内也越来越多的运用,例如刷脸打卡,刷脸APP,身份识别,人脸门禁等。本文将为大家介绍Android基于虹软(ArcSoft)实现人脸识别的demo,快来跟随小编一起学习吧

1、在虹软的开发者中心创建一个自己的应用,将APP_ID与SDK_KEY记录下来,后面会用到。创建完后就可以下载SDK了。

在这里插入图片描述

2、下载完后,就可以根据SDK包里的开发说明文档和代码进行参考和学习。以下是开发说明文档中的SDK包结构的截图。

在这里插入图片描述

3、创建一个空项目,将SDK包里的.jar文件和.so文件复制到该项目的如下包下。接下来的配置十分重要,稍微没处理一个,就是一个头大的bug。

在这里插入图片描述

4、“在app里的build.gradle” 第一个红框原本是androidx的,与support是不兼容的,所以要改,因此,整个项目用到androidx的地方都需要改。第二个红框是ndk,加了这个才能找到刚才复制进去的.so文件。第三个红框也要改成如下。下面的dependencies要注意把androidx的改掉。

在这里插入图片描述

在这里插入图片描述

5、“在整个项目里的build.gradle” 记得加上jcenter()。

在这里插入图片描述

6、在gradle.properties里可能会有androidx的东西,也要删掉。

在这里插入图片描述

7、在AndroidManifest.xml中的中添加权限申请,在中添加。

manifest:

<uses-permission android:name="android.permission.CAMERA" />
    <uses-permission android:name="android.permission.READ_PHONE_STATE" />
    <uses-permission android:name="android.permission.INTERNET" />
    <uses-permission android:name="android.permission.READ_EXTERNAL_STORAGE" />
    <uses-permission android:name="android.permission.WRITE_EXTERNAL_STORAGE" />

provider:

<provider
            android:name="android.support.v4.content.FileProvider"
            android:authorities="${applicationId}.provider"
            android:exported="false"
            android:grantUriPermissions="true">
            <meta-data
                android:name="android.support.FILE_PROVIDER_PATHS"
                android:resource="@xml/provider_paths" />
        </provider>

在添加后要在res下创建一个xml包,里面添加一个provider_paths.xml文件,里面的代码如下:

<?xml version="1.0" encoding="utf-8"?>
<paths xmlns:android="http://schemas.android.com/apk/res/android">
    <external-path name="external_files" path="."/>
    <root-path
        name="root_path"
        path="." />
</paths>

8、从SDK包中引入如下功能包模块和BaseActivity,并将common包下的Constants中的APP_ID,SDK_KEY改成刚才所记录下来的内容。

在这里插入图片描述

9、创建3个acvitity,一个是主界面,一个是人脸库的管理界面,一个是人脸识别功能界面。

在这里插入图片描述

10、layout包下需要引入以下5个布局文件。

在这里插入图片描述

11、主界面主要的功能就是激活权限、连接动态库和激活引擎,我通过修改onCreate()和util包下的ConfigUtil.class的代码,让其能够自动激活和自动修改为全方向人脸检查(其他选择好像不能够实现人脸识别)。以下是激活引擎的代码。

public void activeEngine(final View view) {
        if (!libraryExists) {
            Toast.makeText(this, "未找到库文件!", Toast.LENGTH_SHORT).show();
            return;
        }
        if (!checkPermissions(NEEDED_PERMISSIONS)) {
            ActivityCompat.requestPermissions(this, NEEDED_PERMISSIONS, ACTION_REQUEST_PERMISSIONS);
            return;
        }
        if (view != null) {
            view.setClickable(false);
        }
        Observable.create(new ObservableOnSubscribe<Integer>() {
            @Override
            public void subscribe(ObservableEmitter<Integer> emitter) {
                int activeCode = FaceEngine.activeOnline(MainActivity.this, Constants.APP_ID, Constants.SDK_KEY);
                emitter.onNext(activeCode);
            }
        })
                .subscribeOn(Schedulers.io())
                .observeOn(AndroidSchedulers.mainThread())
                .subscribe(new Observer<Integer>() {
                    @Override
                    public void onSubscribe(Disposable d) {

                    }

                    @Override
                    public void onNext(Integer activeCode) {
                        if (activeCode == ErrorInfo.MOK) {
                            Toast.makeText(MainActivity.this, "激活成功!", Toast.LENGTH_SHORT).show();
                        } else if (activeCode == ErrorInfo.MERR_ASF_ALREADY_ACTIVATED){
                            Toast.makeText(MainActivity.this, "已激活!", Toast.LENGTH_SHORT).show();
                        } else {
                            Toast.makeText(MainActivity.this, "激活失败!", Toast.LENGTH_SHORT).show();
                        }

                        if (view != null) {
                            view.setClickable(true);
                        }
                        ActiveFileInfo activeFileInfo = new ActiveFileInfo();
                    }

                    @Override
                    public void onError(Throwable e) {
                        Toast.makeText(MainActivity.this, e.getMessage(), Toast.LENGTH_SHORT).show();
                        if (view != null) {
                            view.setClickable(true);
                        }
                    }

                    @Override
                    public void onComplete() {

                    }
                });

    }

12、人脸识别界面是最复杂的。其中不仅有人脸识别的功能,还有注册人脸和活体检测的功能。

通过手机自带的摄像头来实现人脸识别和活体检测的逻辑:

private void initCamera() {
        DisplayMetrics metrics = new DisplayMetrics();
        getWindowManager().getDefaultDisplay().getMetrics(metrics);

        final FaceListener faceListener = new FaceListener() {
            @Override
            public void onFail(Exception e) {
                Log.e(TAG, "onFail: " + e.getMessage());
            }

            //请求FR的回调
            @Override
            public void onFaceFeatureInfoGet(@Nullable final FaceFeature faceFeature, final Integer requestId, final Integer errorCode) {
                //FR成功
                if (faceFeature != null) {
//                    Log.i(TAG, "onPreview: fr end = " + System.currentTimeMillis() + " trackId = " + requestId);
                    Integer liveness = livenessMap.get(requestId);
                    //不做活体检测的情况,直接搜索
                    if (!livenessDetect) {
                        searchFace(faceFeature, requestId);
                    }
                    //活体检测通过,搜索特征
                    else if (liveness != null && liveness == LivenessInfo.ALIVE) {
                        searchFace(faceFeature, requestId);
                    }
                    //活体检测未出结果,或者非活体,延迟执行该函数
                    else {
                        if (requestFeatureStatusMap.containsKey(requestId)) {
                            Observable.timer(WAIT_LIVENESS_INTERVAL, TimeUnit.MILLISECONDS)
                                    .subscribe(new Observer<Long>() {
                                        Disposable disposable;

                                        @Override
                                        public void onSubscribe(Disposable d) {
                                            disposable = d;
                                            getFeatureDelayedDisposables.add(disposable);
                                        }

                                        @Override
                                        public void onNext(Long aLong) {
                                            onFaceFeatureInfoGet(faceFeature, requestId, errorCode);
                                        }

                                        @Override
                                        public void onError(Throwable e) {

                                        }

                                        @Override
                                        public void onComplete() {
                                            getFeatureDelayedDisposables.remove(disposable);
                                        }
                                    });
                        }
                    }

                }
                //特征提取失败
                else {
                    if (increaseAndGetValue(extractErrorRetryMap, requestId) > MAX_RETRY_TIME) {
                        extractErrorRetryMap.put(requestId, 0);

                        String msg;
                        // 传入的FaceInfo在指定的图像上无法解析人脸,此处使用的是RGB人脸数据,一般是人脸模糊
                        if (errorCode != null && errorCode == ErrorInfo.MERR_FSDK_FACEFEATURE_LOW_CONFIDENCE_LEVEL) {
                            msg = "人脸置信度低!";
                        } else {
                            msg = "ExtractCode:" + errorCode;
                        }
                        faceHelper.setName(requestId, "未通过!");
                        // 在尝试最大次数后,特征提取仍然失败,则认为识别未通过
                        requestFeatureStatusMap.put(requestId, RequestFeatureStatus.FAILED);
                        retryRecognizeDelayed(requestId);
                    } else {
                        requestFeatureStatusMap.put(requestId, RequestFeatureStatus.TO_RETRY);
                    }
                }
            }

            @Override
            public void onFaceLivenessInfoGet(@Nullable LivenessInfo livenessInfo, final Integer requestId, Integer errorCode) {
                if (livenessInfo != null) {
                    int liveness = livenessInfo.getLiveness();
                    livenessMap.put(requestId, liveness);
                    // 非活体,重试
                    if (liveness == LivenessInfo.NOT_ALIVE) {
                        faceHelper.setName(requestId, "未通过!非活体!");
                        // 延迟 FAIL_RETRY_INTERVAL 后,将该人脸状态置为UNKNOWN,帧回调处理时会重新进行活体检测
                        retryLivenessDetectDelayed(requestId);
                    }
                } else {
                    if (increaseAndGetValue(livenessErrorRetryMap, requestId) > MAX_RETRY_TIME) {
                        livenessErrorRetryMap.put(requestId, 0);
                        String msg;
                        // 传入的FaceInfo在指定的图像上无法解析人脸,此处使用的是RGB人脸数据,一般是人脸模糊
                        if (errorCode != null && errorCode == ErrorInfo.MERR_FSDK_FACEFEATURE_LOW_CONFIDENCE_LEVEL) {
                            msg = "人脸置信度低!";
                        } else {
                            msg = "ProcessCode:" + errorCode;
                        }
                        faceHelper.setName(requestId, "未通过!");
                        retryLivenessDetectDelayed(requestId);
                    } else {
                        livenessMap.put(requestId, LivenessInfo.UNKNOWN);
                    }
                }
            }


        };


        CameraListener cameraListener = new CameraListener() {
            @Override
            public void onCameraOpened(Camera camera, int cameraId, int displayOrientation, boolean isMirror) {
                Camera.Size lastPreviewSize = previewSize;
                previewSize = camera.getParameters().getPreviewSize();
                drawHelper = new DrawHelper(previewSize.width, previewSize.height, previewView.getWidth(), previewView.getHeight(), displayOrientation
                        , cameraId, isMirror, false, false);
                Log.i(TAG, "onCameraOpened: " + drawHelper.toString());
                // 切换相机的时候可能会导致预览尺寸发生变化
                if (faceHelper == null ||
                        lastPreviewSize == null ||
                        lastPreviewSize.width != previewSize.width || lastPreviewSize.height != previewSize.height) {
                    Integer trackedFaceCount = null;
                    // 记录切换时的人脸序号
                    if (faceHelper != null) {
                        trackedFaceCount = faceHelper.getTrackedFaceCount();
                        faceHelper.release();
                    }
                    faceHelper = new FaceHelper.Builder()
                            .ftEngine(ftEngine)
                            .frEngine(frEngine)
                            .flEngine(flEngine)
                            .frQueueSize(MAX_DETECT_NUM)
                            .flQueueSize(MAX_DETECT_NUM)
                            .previewSize(previewSize)
                            .faceListener(faceListener)
                            .trackedFaceCount(trackedFaceCount == null ? ConfigUtil.getTrackedFaceCount(FaceRegisterAndRecognise.this.getApplicationContext()) : trackedFaceCount)
                            .build();
                }
            }


            @Override
            public void onPreview(final byte[] nv21, Camera camera) {
                if (faceRectView != null) {
                    faceRectView.clearFaceInfo();
                }
                List<FacePreviewInfo> facePreviewInfoList = faceHelper.onPreviewFrame(nv21);
                if (facePreviewInfoList != null && faceRectView != null && drawHelper != null) {
                    drawPreviewInfo(facePreviewInfoList);
                }
                registerFace(nv21, facePreviewInfoList);
                clearLeftFace(facePreviewInfoList);

                if (facePreviewInfoList != null && facePreviewInfoList.size() > 0 && previewSize != null) {
                    for (int i = 0; i < facePreviewInfoList.size(); i++) {
                        Integer status = requestFeatureStatusMap.get(facePreviewInfoList.get(i).getTrackId());
                        /**
                         * 在活体检测开启,在人脸识别状态不为成功或人脸活体状态不为处理中(ANALYZING)且不为处理完成(ALIVE、NOT_ALIVE)时重新进行活体检测
                         */
                        if (livenessDetect && (status == null || status != RequestFeatureStatus.SUCCEED)) {
                            Integer liveness = livenessMap.get(facePreviewInfoList.get(i).getTrackId());
                            if (liveness == null
                                    || (liveness != LivenessInfo.ALIVE && liveness != LivenessInfo.NOT_ALIVE && liveness != RequestLivenessStatus.ANALYZING)) {
                                livenessMap.put(facePreviewInfoList.get(i).getTrackId(), RequestLivenessStatus.ANALYZING);
                                faceHelper.requestFaceLiveness(nv21, facePreviewInfoList.get(i).getFaceInfo(), previewSize.width, previewSize.height, FaceEngine.CP_PAF_NV21, facePreviewInfoList.get(i).getTrackId(), LivenessType.RGB);
                            }
                        }
                        /**
                         * 对于每个人脸,若状态为空或者为失败,则请求特征提取(可根据需要添加其他判断以限制特征提取次数),
                         * 特征提取回传的人脸特征结果在{@link FaceListener#onFaceFeatureInfoGet(FaceFeature, Integer, Integer)}中回传
                         */
                        if (status == null
                                || status == RequestFeatureStatus.TO_RETRY) {
                            requestFeatureStatusMap.put(facePreviewInfoList.get(i).getTrackId(), RequestFeatureStatus.SEARCHING);
                            faceHelper.requestFaceFeature(nv21, facePreviewInfoList.get(i).getFaceInfo(), previewSize.width, previewSize.height, FaceEngine.CP_PAF_NV21, facePreviewInfoList.get(i).getTrackId());
//                            Log.i(TAG, "onPreview: fr start = " + System.currentTimeMillis() + " trackId = " + facePreviewInfoList.get(i).getTrackedFaceCount());
                        }
                    }
                }
            }

            @Override
            public void onCameraClosed() {
                Log.i(TAG, "onCameraClosed: ");
            }

            @Override
            public void onCameraError(Exception e) {
                Log.i(TAG, "onCameraError: " + e.getMessage());
            }

            @Override
            public void onCameraConfigurationChanged(int cameraID, int displayOrientation) {
                if (drawHelper != null) {
                    drawHelper.setCameraDisplayOrientation(displayOrientation);
                }
                Log.i(TAG, "onCameraConfigurationChanged: " + cameraID + "  " + displayOrientation);
            }
        };

        cameraHelper = new CameraHelper.Builder()
                .previewViewSize(new Point(previewView.getMeasuredWidth(), previewView.getMeasuredHeight()))
                .rotation(getWindowManager().getDefaultDisplay().getRotation())
                .specificCameraId(rgbCameraID != null ? rgbCameraID : Camera.CameraInfo.CAMERA_FACING_FRONT)
                .isMirror(false)
                .previewOn(previewView)
                .cameraListener(cameraListener)
                .build();
        cameraHelper.init();
        cameraHelper.start();
    }

注册人脸的逻辑:

private void registerFace(final byte[] nv21, final List<FacePreviewInfo> facePreviewInfoList) {
        if (registerStatus == REGISTER_STATUS_READY && facePreviewInfoList != null && facePreviewInfoList.size() > 0) {
            registerStatus = REGISTER_STATUS_PROCESSING;
            Observable.create(new ObservableOnSubscribe<Boolean>() {
                @Override
                public void subscribe(ObservableEmitter<Boolean> emitter) {

                    boolean success = FaceServer.getInstance().registerNv21(FaceRegisterAndRecognise.this, nv21.clone(), previewSize.width, previewSize.height,
                            facePreviewInfoList.get(0).getFaceInfo(), "registered " + faceHelper.getTrackedFaceCount());
                    emitter.onNext(success);
                }
            })
                    .subscribeOn(Schedulers.computation())
                    .observeOn(AndroidSchedulers.mainThread())
                    .subscribe(new Observer<Boolean>() {
                        @Override
                        public void onSubscribe(Disposable d) {

                        }

                        @Override
                        public void onNext(Boolean success) {
                            String result = success ? "register success!" : "register failed!";
                            showToast(result);
                            registerStatus = REGISTER_STATUS_DONE;
                        }

                        @Override
                        public void onError(Throwable e) {
                            e.printStackTrace();
                            showToast("register failed!");
                            registerStatus = REGISTER_STATUS_DONE;
                        }

                        @Override
                        public void onComplete() {

                        }
                    });
        }
    }

13、人脸库的管理界面。

public class FaceLibs extends BaseActivity {
    private ExecutorService executorService;
    private TextView textView;

    private TextView tvNotificationRegisterResult;

    ProgressDialog progressDialog = null;
    private static final int ACTION_REQUEST_PERMISSIONS = 0x001;
    private static String[] NEEDED_PERMISSIONS = new String[]{
            Manifest.permission.READ_EXTERNAL_STORAGE,
            Manifest.permission.WRITE_EXTERNAL_STORAGE
    };

    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_face_libs);
        getWindow().addFlags(WindowManager.LayoutParams.FLAG_KEEP_SCREEN_ON);
        executorService = Executors.newSingleThreadExecutor();
        tvNotificationRegisterResult = findViewById(R.id.notification_register_result);
        progressDialog = new ProgressDialog(this);
        int faceLibNum = FaceServer.getInstance().getFaceNumber(this);
        textView = findViewById(R.id.number);
        textView.setText(faceLibNum + "");
        FaceServer.getInstance().init(this);
    }

    @Override
    protected void onDestroy() {
        if (executorService != null && !executorService.isShutdown()) {
            executorService.shutdownNow();
        }
        if (progressDialog != null && progressDialog.isShowing()) {
            progressDialog.dismiss();
        }

        FaceServer.getInstance().unInit();
        super.onDestroy();
    }

    @Override
    void afterRequestPermission(int requestCode, boolean isAllGranted) {

    }

    public void clearFaces(View view) {
        int faceNum = FaceServer.getInstance().getFaceNumber(this);
        if (faceNum == 0) {
            showToast("人脸库已空!");
        } else {
            AlertDialog dialog = new AlertDialog.Builder(this)
                    .setTitle("通知")
                    .setMessage("确定要删除" + faceNum + "个人脸吗?")
                    .setPositiveButton("确定", new DialogInterface.OnClickListener() {
                        @Override
                        public void onClick(DialogInterface dialog, int which) {
                            int deleteCount = FaceServer.getInstance().clearAllFaces(FaceLibs.this);
                            showToast(deleteCount + "个人脸已删除!");
                            textView.setText("0");
                        }
                    })
                    .setNegativeButton("取消", null)
                    .create();
            dialog.show();
        }
    }
}

14、以上就是大体的介绍,还有一些小的细枝末节需要同志们动手实操一下。下面就来看看实现的效果。

主界面:

在这里插入图片描述

注册成功并通过识别:

在这里插入图片描述

通过手机照片识别出不是活体:

在这里插入图片描述

清理人脸库:

在这里插入图片描述

以上就是Android基于虹软(ArcSoft)实现人脸识别的详细内容,更多关于Android人脸识别的资料请关注脚本之家其它相关文章!

您可能感兴趣的文章:
阅读全文