簡體   English   中英

Android Face檢測僅適用於不使用SD卡圖像的drawable

[英]Android Face detection only works with drawables not with images from SD card

所以我有代碼在任何給定的圖像文件中檢測多達10個面孔並返回給我信息,如眼睛的位置和其他類似的東西。 因此,當我告訴它使用存儲在我的項目的資源的drawable文件夾中的圖像文件時,它工作得很好。 但是,當我有它嘗試從我從SD卡導入的位圖找到面孔時,它不會找到任何面孔。 但這些都是完全相同的圖像。 有任何想法嗎? 我的代碼如下:

編輯:經過進一步檢查后,我發現當我插入這行代碼時System.out.println("Row Bytes: " + sourceImage.getRowBytes()); 我得到的drawable是352和SD卡圖像是704.我認為這意味着drawable在.apk文件中被壓縮,但SD卡圖像顯然不是。 不確定這是否會影響任何事情。

 public class FaceView extends View {
           private static final int NUM_FACES = 10; // max is 64
           private static final boolean DEBUG = true;

           private FaceDetector arrayFaces;
           private FaceDetector.Face getAllFaces[] = new FaceDetector.Face[NUM_FACES];
           private FaceDetector.Face getFace = null;

           private PointF eyesMidPts[] = new PointF[NUM_FACES];
           private float  eyesDistance[] = new float[NUM_FACES];

           private Bitmap sourceImage;

           private Paint tmpPaint = new Paint(Paint.ANTI_ALIAS_FLAG);
           private Paint pOuterBullsEye = new Paint(Paint.ANTI_ALIAS_FLAG);
           private Paint pInnerBullsEye = new Paint(Paint.ANTI_ALIAS_FLAG);

           private int picWidth, picHeight;
           private float xRatio, yRatio;

           public FaceView(Context context) {
                   super(context);

                   pInnerBullsEye.setStyle(Paint.Style.FILL);
                   pInnerBullsEye.setColor(Color.RED);

                   pOuterBullsEye.setStyle(Paint.Style.STROKE);
                   pOuterBullsEye.setColor(Color.RED);

                   tmpPaint.setStyle(Paint.Style.STROKE);
                   tmpPaint.setTextAlign(Paint.Align.CENTER);

                   BitmapFactory.Options bfo = new BitmapFactory.Options();
                   bfo.inPreferredConfig = Bitmap.Config.RGB_565;

                   //********This code imports the image from the SD card which does not work
                   String imageInSD = Environment.getExternalStorageDirectory()
                        .getAbsolutePath() + "/testfolder/" + "face1" + ".png";

                   Bitmap sourceImage = BitmapFactory.decodeFile(imageInSD,bfo);

                   //**********This code uses an image in the projects drawable folder, this code works.
                   sourceImage = BitmapFactory.decodeResource( getResources() ,R.drawable.face1, bfo);

                   picWidth = sourceImage.getWidth();
                   picHeight = sourceImage.getHeight();

                   arrayFaces = new FaceDetector( picWidth, picHeight, NUM_FACES );
                   arrayFaces.findFaces(sourceImage, getAllFaces);

                   for (int i = 0; i < getAllFaces.length; i++)
                   {
                           getFace = getAllFaces[i];
                           try {
                                   PointF eyesMP = new PointF();
                                   getFace.getMidPoint(eyesMP);
                                   eyesDistance[i] = getFace.eyesDistance();
                                   eyesMidPts[i] = eyesMP;

                                   if (DEBUG)
                                   {
                                           Log.i("Face",
                                                   i +  " " + getFace.confidence() + " " + getFace.eyesDistance() + " "
                                                   + "Pose: ("+ getFace.pose(FaceDetector.Face.EULER_X) + ","
                                                   + getFace.pose(FaceDetector.Face.EULER_Y) + ","
                                                   + getFace.pose(FaceDetector.Face.EULER_Z) + ")"
                                                   + "Eyes Midpoint: ("+eyesMidPts[i].x + "," + eyesMidPts[i].y +")"
                                           );
                                   }
                           }
                           catch (Exception e)
                           {
                                   if (DEBUG) Log.e("Face", i + " is null");
                           }

                   }


           }

           @Override
           protected void onDraw(Canvas canvas)
           {
                   xRatio = getWidth()*1.0f / picWidth;
                   yRatio = getHeight()*1.0f / picHeight;
                   canvas.drawBitmap( sourceImage, null , new Rect(0,0,getWidth(),getHeight()),tmpPaint);
                   for (int i = 0; i < eyesMidPts.length; i++)
                   {
                           if (eyesMidPts[i] != null)
                           {
                                   pOuterBullsEye.setStrokeWidth(eyesDistance[i] /6);
                                   canvas.drawCircle(eyesMidPts[i].x*xRatio, eyesMidPts[i].y*yRatio, eyesDistance[i] / 2 , pOuterBullsEye);
                                   canvas.drawCircle(eyesMidPts[i].x*xRatio, eyesMidPts[i].y*yRatio, eyesDistance[i] / 6 , pInnerBullsEye);
                           }
                   }


           }  

}

好吧,我相信我知道你的問題在這里。 設備無法將圖像渲染為位圖圖像,因為它位於外部存儲器中。 面部識別工作只是沒有進入畫布。 我的xoom所有設備都有渲染限制(2048x2048)我在這里找到 將圖像作為資源添加時它​​起作用的原因是因為你的文件縮小了,因為它構建了.apk(說實話我不知道為什么會這樣做,但是我留下了一些println用於測試,其他人可以更好地回答這個問題。 無論如何,我只是在你的代碼尋找面之后,在它嘗試將位圖渲染到畫布之前,除以2來縮放位圖。 現在一切似乎都很好。 您可能需要調整面部指示器但功能正常。 我希望這有幫助。

public class FaceView extends View {
private static final int NUM_FACES = 1; // max is 64
private static final boolean DEBUG = true;

private FaceDetector arrayFaces;
private FaceDetector.Face getAllFaces[] = new FaceDetector.Face[NUM_FACES];
private FaceDetector.Face getFace = null;

private PointF eyesMidPts[] = new PointF[NUM_FACES];
private float  eyesDistance[] = new float[NUM_FACES];

private Bitmap sourceImage;

private Paint tmpPaint = new Paint(Paint.ANTI_ALIAS_FLAG);
private Paint pOuterBullsEye = new Paint(Paint.ANTI_ALIAS_FLAG);
private Paint pInnerBullsEye = new Paint(Paint.ANTI_ALIAS_FLAG);

private int picWidth, picHeight;
private float xRatio, yRatio;

public FaceView(Context context) {
        super(context);

        pInnerBullsEye.setStyle(Paint.Style.FILL);
        pInnerBullsEye.setColor(Color.RED);

        pOuterBullsEye.setStyle(Paint.Style.STROKE);
        pOuterBullsEye.setColor(Color.RED);

        tmpPaint.setStyle(Paint.Style.STROKE);
        tmpPaint.setTextAlign(Paint.Align.CENTER);

        BitmapFactory.Options bfo = new BitmapFactory.Options();
        bfo.inPreferredConfig = Bitmap.Config.RGB_565;

        //********This code imports the image from the SD card which does not work
        String imageInSD = Environment.getExternalStorageDirectory().getAbsolutePath() + "/face1" + ".jpg";

        System.out.println(imageInSD);

        sourceImage = BitmapFactory.decodeFile(imageInSD, bfo);

        //Bitmap sourceImage;// = BitmapFactory.decodeFile(imageInSD,bfo);


        //**********This code uses an image in the projects drawable folder, this code works.
        //sourceImage = BitmapFactory.decodeResource( getResources() ,R.drawable.face1, bfo);

        picWidth = sourceImage.getWidth();
        picHeight = sourceImage.getHeight(); 

        System.out.println(picWidth + "x" + picHeight);

        arrayFaces = new FaceDetector( picWidth, picHeight, NUM_FACES );
        arrayFaces.findFaces(sourceImage, getAllFaces);

        sourceImage = Bitmap.createScaledBitmap (sourceImage, picWidth/2, picHeight/2, false);

        for (int i = 0; i < getAllFaces.length; i++)
        {
                getFace = getAllFaces[i];
                try {
                        PointF eyesMP = new PointF();
                        getFace.getMidPoint(eyesMP);
                        eyesDistance[i] = getFace.eyesDistance();
                        eyesMidPts[i] = eyesMP;

                        if (DEBUG)
                        {
                                Log.i("Face",
                                        i +  " " + getFace.confidence() + " " + getFace.eyesDistance() + " "
                                        + "Pose: ("+ getFace.pose(FaceDetector.Face.EULER_X) + ","
                                        + getFace.pose(FaceDetector.Face.EULER_Y) + ","
                                        + getFace.pose(FaceDetector.Face.EULER_Z) + ")"
                                        + "Eyes Midpoint: ("+eyesMidPts[i].x + "," + eyesMidPts[i].y +")"
                                );
                        }
                }
                catch (Exception e)
                {
                        if (DEBUG) Log.e("Face", i + " is null");
                }

        }


}

@Override
protected void onDraw(Canvas canvas)
{
        xRatio = getWidth()*1.0f / picWidth; 
        yRatio = getHeight()*1.0f / picHeight;
        canvas.drawBitmap( sourceImage, null , new Rect(0,0,getWidth(),getHeight()),tmpPaint);
        for (int i = 0; i < eyesMidPts.length; i++)
        {
                if (eyesMidPts[i] != null)
                {
                        pOuterBullsEye.setStrokeWidth(eyesDistance[i] /6);
                        canvas.drawCircle(eyesMidPts[i].x*xRatio, eyesMidPts[i].y*yRatio, eyesDistance[i] / 2 , pOuterBullsEye);
                        canvas.drawCircle(eyesMidPts[i].x*xRatio, eyesMidPts[i].y*yRatio, eyesDistance[i] / 6 , pInnerBullsEye);
                }
        }


}

}

事實證明,相機拍攝的照片會保存為PNG文件,只有在使用JPG文件時才可以從SD卡成功進行人臉檢測。 只需將文件轉換為JPG,它就可以正常工作。

暫無
暫無

聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.

 
粵ICP備18138465號  © 2020-2024 STACKOOM.COM