前段时间，公司项目需要运用人脸识别，我用的AVFoundation、CoreImage的CIFaceFeature来实现，然后一顿操作完成过后，发现人脸识别时在6s上CPU使用率达到了70%，在6以下的真机上运行CPU直接到150%+。。。卡到爆炸💥
然后代码定位在人脸识别的代理上
#pragma mark - <AVCaptureVideoDataOutputSampleBufferDelegate>
- (void)captureOutput:(AVCaptureOutput *)captureOutput didOutputSampleBuffer:(CMSampleBufferRef)sampleBuffer fromConnection:(AVCaptureConnection *)connection {
CVImageBufferRef imageBuffer = CMSampleBufferGetImageBuffer(sampleBuffer);
CVPixelBufferLockBaseAddress(imageBuffer,0);
uint8_t *baseAddress = (uint8_t *)CVPixelBufferGetBaseAddress(imageBuffer);
size_t bytesPerRow = CVPixelBufferGetBytesPerRow(imageBuffer);
size_t width = CVPixelBufferGetWidth(imageBuffer);
size_t height = CVPixelBufferGetHeight(imageBuffer);
CGColorSpaceRef colorSpace = CGColorSpaceCreateDeviceRGB();
CGContextRef newContext = CGBitmapContextCreate(baseAddress,
width, height, 8, bytesPerRow, colorSpace,
kCGBitmapByteOrder32Little | kCGImageAlphaPremultipliedFirst);
CGImageRef newImage = CGBitmapContextCreateImage(newContext);
CGContextRelease(newContext);
CGColorSpaceRelease(colorSpace);
UIImage *image= [UIImage imageWithCGImage:newImage scale:1 orientation:UIImageOrientationLeftMirrored];
CGImageRelease(newImage);
[self performSelectorOnMainThread:@selector(detectForFacesInUIImage:)
withObject: (id) image waitUntilDone:NO];
CVPixelBufferUnlockBaseAddress(imageBuffer,0);
}
一般视频录制是30帧，机子每秒都需要处理那么多张图片根本吃不消，特别是CPU比较老的机型。
我的解决方法是：
1.限制摄像头的帧数
[videoDevice setActiveVideoMaxFrameDuration:CMTimeMake(1, 25)];
限制最大帧数 每秒25帧
2.每5帧取一张图片来进行处理
fpsCount_ ++;
if (fpsCount_%5 != 1) {
return;
}
先定义一个全局变量fpsCount_，然后在didOutputSampleBuffer:方法开头进行控制。
最后，终于在老机型上成功流畅运行，�运行时CPU使用率只有百分之二三十。
