如何将 MLKit Vision3DPoint 转换为 UV 坐标 目标详情地标位置示例预期结果代码

如何解决如何将 MLKit Vision3DPoint 转换为 UV 坐标 目标详情地标位置示例预期结果代码

目标

• 转换 MLKit Vision3DPoint 到 UV坐标或AVFoundation坐标

详情

我尝试通过遵循 these steps 来使用 Google 的 MLKit 来检测人体姿势，但结果是在 3D 空间中

地标位置示例

Position:  x: 1175.290893554688,y: -3.273239135742188,z: 2.268885850906372
Position:  x: 1283.07568359375,y: 18.44030952453613,z: -266.7630615234375
Position:  x: 4367.9072265625,y: 687.6997680664062,z: -358.7154846191406
Position:  x: 3602.029052734375,y: 621.983642578125,z: -1123.94580078125
...

但我想要的答案是图像上的每个点（UV坐标）？

预期结果

// Example
UV:  u: 0.1,v: 0.2,Screen: x: 100,y: 200
...

代码

func captureOutput(_ output: AVCaptureOutput,didOutput sampleBuffer: CMSampleBuffer,from connection: AVCaptureConnection) {
    
    // Prepare Input
    let image = VisionImage(buffer: sampleBuffer)
    image.orientation = imageOrientation(deviceOrientation: UIDevice.current.orientation,cameraPosition: camera.position)
    
    // To detect objects asynchronously
    poseDetector.process(image) { [uNowned self] (detectedPoses,error) in
        
        guard error == nil else { return }
        
        guard let detectedPoses = detectedPoses,!detectedPoses.isEmpty else { return }
        
        // Success. Get pose landmarks here.
        
        for pose in detectedPoses {
            let landmarks = pose.landmarks
            for landmark in landmarks {

                // This will return a 3D point
                // Example: x: 1175.290893554688,z: 2.268885850906372
                let position = landmark.position

                // pointView.points needs 2D Point (UV Coordinate)
                // Example: x: 0.12,y: 0.78 (u,v)
                owner.pointView.points.append((CGPoint(x: position.x,y: position.y),UIColor.red))
            }
        }
        owner.pointView.setNeedsdisplay()
    }
    
}