Face Detection not working in Portrait Mode

[Qirtas]

Hi, I'm using Samsung Galaxy j5 (2016) model. On this face detection is not working in portrait mode but working fine in landscape only.

I've tested on Nexus 5, Galaxy TAB A and LG G Pro devices and on all these devices your library is working perfectly for both orientations. Can you help me please?

[VISHALBHAGWAT88]

fetching same issue !

[zhangxiaang]

I find the resolution of this Issue:

fdet.findFaces(bmp, fullResults);

We should ensure the width and height of bitmap are even. It works for me.

`gray8toRGB32(grayBuff, previewWidth, previewHeight, rgbs); Bitmap bitmap = Bitmap.createBitmap(rgbs, previewWidth, previewHeight, Bitmap.Config.RGB_565);

        if (w % 2 == 1) {
            w -= 1;
        }
        if (h % 2 == 1) {
            h -= 1;
        }//here
        Bitmap bmp = Bitmap.createScaledBitmap(bitmap, w, h, false);

        float xScale = (float) previewWidth / (float) prevSettingWidth;
        float yScale = (float) previewHeight / (float) h;

        Camera.CameraInfo info = new Camera.CameraInfo();
        Camera.getCameraInfo(cameraId, info);
        int rotate = mDisplayOrientation;
        if (info.facing == Camera.CameraInfo.CAMERA_FACING_FRONT && mDisplayRotation % 180 == 0) {
            if (rotate + 180 > 360) {
                rotate = rotate - 180;
            } else {
                rotate = rotate + 180;
            }
        }

        switch (rotate) {
            case 90:
                bmp = ImageUtils.rotate(bmp, 90);
                xScale = (float) previewHeight / bmp.getWidth();
                yScale = (float) previewWidth / bmp.getHeight();
                break;
            case 180:
                bmp = ImageUtils.rotate(bmp, 180);
                break;
            case 270:
                bmp = ImageUtils.rotate(bmp, 270);
                xScale = (float) previewHeight / (float) h;
                yScale = (float) previewWidth / (float) prevSettingWidth;
                break;
            default:
                break;
        }

        fdet = new FaceDetector(bmp.getWidth(), bmp.getHeight(), MAX_FACE);
        FaceDetector.Face[] fullResults = new FaceDetector.Face[MAX_FACE];
        fdet.findFaces(bmp, fullResults);`

[VISHALBHAGWAT88]

Tysm for helping ..

[zhangxiaang]

But I find this project cost too much CPU resources (15~20% CPU), I'm trying to find a high performance method to detect faces。