ios - 由于缩放，使用 ImageView 进行人脸检测和缩放效果不佳？

Question

再一次，我很接近，但没有香蕉。

我正在尝试学习一些关于面部识别的教程。使用以下代码几乎就可以了，但我认为在缩放和在面部周围放置 UIImageView 边框方面我缺少一些东西。

我在照片库中的照片大小不同(出于某些无法解释的原因)，所以我想 CIDetector 正在寻找人脸，我正在应用 CGAffineTransforms 等，并尝试将它们放在 UIImageView 中。但是，正如您从图像(也在下方)中看到的那样，它没有绘制在正确的位置。

UIImageView 为 280x500 并设置为 Scale to Fill。

任何帮助弄清楚正在发生的事情都会很棒!

-(void)detectFaces {

    CIContext *context = [CIContext contextWithOptions:nil];
    CIImage *image = [CIImage imageWithCGImage:_imagePhotoChosen.image.CGImage options:nil];
    CIDetector *detector = [CIDetector detectorOfType:CIDetectorTypeFace context:context options:@{CIDetectorAccuracy : CIDetectorAccuracyHigh}];
    CGAffineTransform transform = CGAffineTransformMakeScale(1, -1);
    transform = CGAffineTransformTranslate(transform, 0, -_imagePhotoChosen.image.size.height);
    NSArray *features = [detector featuresInImage:image];
    NSLog(@"I have found %lu faces", (long unsigned)features.count);
    for (CIFaceFeature *faceFeature in features)
    {   
        const CGRect faceRect = CGRectApplyAffineTransform(faceFeature.bounds, transform);
        NSLog(@"I have the original frame as: %@", NSStringFromCGRect(faceRect));
        const CGFloat scaleWidth = _imagePhotoChosen.frame.size.width/_imagePhotoChosen.image.size.width;
        const CGFloat scaleHeight = _imagePhotoChosen.frame.size.height/_imagePhotoChosen.image.size.height;

        CGRect faceFrame = CGRectMake(faceRect.origin.x * scaleWidth, faceRect.origin.y * scaleHeight, faceRect.size.width * scaleWidth, faceRect.size.height * scaleHeight);

        UIView *faceView = [[UIView alloc] initWithFrame:faceFrame];
        NSLog(@"I have the bounds as: %@", NSStringFromCGRect(faceFrame));
        faceView.layer.borderColor = [[UIColor redColor] CGColor];
        faceView.layer.borderWidth = 1.0f;

        [self.view addSubview:faceView];
    }

}

-(void) imagePickerController:(UIImagePickerController *)picker didFinishPickingMediaWithInfo:(NSDictionary *)info{

    _imagePhotoChosen.image = info[UIImagePickerControllerOriginalImage];
    //[_imagePhotoChosen sizeToFit];

    [self.view addSubview:_viewChosenPhoto];
    [picker dismissViewControllerAnimated:YES completion:nil];
    [self detectFaces];

}

我已经留在 NSLog 语句中，因为我一直在尝试计算数学是否错误，但似乎无法确定它是否正确!而且我也是一名数学老师......叹息....

再次感谢您为我指明正确方向所做的一切。

更新

作为对想知道我是如何解决它的人的回应......这对我来说真的是一个愚蠢的错误。

我将 subview 添加到主窗口，而不是 UIImageView。因此我删除了这一行:

[self.view addSubview:faceView];

并将其替换为:

[_imagePhotoChosen addSubview:faceView];

这样就可以将框架放置在正确的位置。公认的解决方案为我提供了线索!所以，更新后的代码(从那以后我又做了一些改动)变成了:

-(void)detectFaces:(UIImage *)selectedImage {

    _imagePhotoChosen.image = selectedImage;

    CIImage *image = [CIImage imageWithCGImage:selectedImage.CGImage options:nil];

    CIDetector *detector = [CIDetector detectorOfType:CIDetectorTypeFace context:nil options:@{CIDetectorAccuracy : CIDetectorAccuracyHigh}];
    CGAffineTransform transform = CGAffineTransformMakeScale(1, -1);
    transform = CGAffineTransformTranslate(transform, 0, -selectedImage.size.height);
    NSArray *features = [detector featuresInImage:image];
    int i = 0;
    for (CIFaceFeature *faceFeature in features)
    {

        const CGRect faceRect = CGRectApplyAffineTransform(faceFeature.bounds, transform);

        const CGFloat scaleWidth = _imagePhotoChosen.frame.size.width/_imagePhotoChosen.image.size.width;
        const CGFloat scaleHeight = _imagePhotoChosen.frame.size.height/_imagePhotoChosen.image.size.height;

        CGRect faceFrame = CGRectMake(faceRect.origin.x * scaleWidth, faceRect.origin.y * scaleHeight, faceRect.size.width * scaleWidth, faceRect.size.height * scaleHeight);

        UIView *faceView = [[UIView alloc] initWithFrame:faceFrame];
        faceView.layer.borderColor = [[UIColor redColor] CGColor];
        faceView.layer.borderWidth = 1.0f;
        faceView.tag = i;

        UITapGestureRecognizer *selectPhotoTap = [[UITapGestureRecognizer alloc] initWithTarget:self action:@selector(selectPhoto)];
        selectPhotoTap.numberOfTapsRequired = 1;
        selectPhotoTap.numberOfTouchesRequired = 1;
        [faceView addGestureRecognizer:selectPhotoTap];

        [_imagePhotoChosen addSubview:faceView];
        i++;
    }

}

Answer 1

其实你做的是完全正确的，把这一行换成

    CGRect faceFrame = CGRectMake(_imagePhotoChosen.frame.origin.x+ faceRect.origin.x * scaleWidth,_imagePhotoChosen.frame.origin.y+ faceRect.origin.y * scaleHeight, faceRect.size.width * scaleWidth, faceRect.size.height * scaleHeight);