MediaPipeUnityPlugin: HandLandmark Positions Don't Map Correctly

Plugin Version or Commit ID

v0.10.1

Unity Version

2021.3.3f1

Your Host OS

macOS Monterey 12.2

Target Platform

Android

Description

Hi, I am using this plugin with AR Foundation and I have an issue about matching the hand landmarks provided by Mediapipe graph. The landmarks come in the form of (0, 0) and (1, 1). I am trying to convert these 21 landmarks following the tutorial, by using rect.GetPoint(). But landmarks don’t match correctly, they shift a little bit. I cannot solve this issue, and don’t know why this is happening. Could you please help me?

Here is the result in my phone:

Here is my code:

`using System; using System.Collections; using System.Collections.Generic;

using Unity.Collections; using Unity.Collections.LowLevel.Unsafe; using UnityEngine; using UnityEngine.UI; using UnityEngine.XR.ARFoundation; using UnityEngine.XR.ARSubsystems;

using Mediapipe; using Mediapipe.Unity; using Mediapipe.Unity.CoordinateSystem;

using Stopwatch = System.Diagnostics.Stopwatch;

public class HandTrackingAR : MonoBehaviour { private enum ModelComplexity { Lite = 0, Full = 1, }

private enum MaxNumberHands
{
    One = 1,
    Two = 2,
}

[SerializeField] private TextAsset _gpuConfig;
[SerializeField] private ModelComplexity _modelComplexity = ModelComplexity.Full;
[SerializeField] private MaxNumberHands _maxNumHands = MaxNumberHands.One;
[SerializeField] private RawImage _screen;
[SerializeField] private ARCameraManager _cameraManager;
[SerializeField] private GameObject _handPoint;

private CalculatorGraph _graph;
private static UnityEngine.Rect _screenRect;
private GameObject[] hand;
private Stopwatch _stopwatch;
private ResourceManager _resourceManager;
private GpuResources _gpuResources;
private NativeArray<byte> _buffer;
private OutputStream<NormalizedLandmarkListVectorPacket, List<NormalizedLandmarkList>> _handLandmarkStream;

private IEnumerator Start()
{
    Glog.Logtostderr = true;
    Glog.Minloglevel = 0;
    Glog.V = 3;
    Protobuf.SetLogHandler(Protobuf.DefaultLogHandler);

    _cameraManager.frameReceived += OnCameraFrameReceived;
    _gpuResources = GpuResources.Create().Value();
    _stopwatch = new Stopwatch();

    hand = new GameObject[21];
    for (var i = 0; i < 21; i++)
    {
        hand[i] = Instantiate(_handPoint, _screen.transform);
    }

    _resourceManager = new StreamingAssetsResourceManager();
    if (_modelComplexity == ModelComplexity.Lite)
    {
        yield return _resourceManager.PrepareAssetAsync("hand_landmark_lite.bytes");
        yield return _resourceManager.PrepareAssetAsync("hand_recrop.bytes");
        yield return _resourceManager.PrepareAssetAsync("handedness.txt");
        yield return _resourceManager.PrepareAssetAsync("palm_detection_lite.bytes");
    }
    else
    {
        yield return _resourceManager.PrepareAssetAsync("hand_landmark_full.bytes");
        yield return _resourceManager.PrepareAssetAsync("hand_recrop.bytes");
        yield return _resourceManager.PrepareAssetAsync("handedness.txt");
        yield return _resourceManager.PrepareAssetAsync("palm_detection_full.bytes");
    }

    _graph = new CalculatorGraph(_gpuConfig.text);
    _graph.SetGpuResources(_gpuResources).AssertOk();
    _screenRect = _screen.GetComponent<RectTransform>().rect;

    _handLandmarkStream = new OutputStream<NormalizedLandmarkListVectorPacket, List<NormalizedLandmarkList>>(_graph, "hand_landmarks");
    _handLandmarkStream.StartPolling().AssertOk();

    var sidePacket = new SidePacket();
    sidePacket.Emplace("model_complexity", new IntPacket((int)_modelComplexity));
    sidePacket.Emplace("num_hands", new IntPacket((int)_maxNumHands));
    sidePacket.Emplace("input_rotation", new IntPacket(270));
    sidePacket.Emplace("input_horizontally_flipped", new BoolPacket(true));
    sidePacket.Emplace("input_vertically_flipped", new BoolPacket(true));

    _graph.StartRun(sidePacket).AssertOk();
    _stopwatch.Start();
}

private unsafe void OnCameraFrameReceived(ARCameraFrameEventArgs eventArgs)
{
    if (_cameraManager.TryAcquireLatestCpuImage(out XRCpuImage image))
    {
        alocBuffer(image);
        var conversionParams = new XRCpuImage.ConversionParams(image, TextureFormat.RGBA32);
        var ptr = (IntPtr)NativeArrayUnsafeUtility.GetUnsafePtr(_buffer);
        image.Convert(conversionParams, ptr, _buffer.Length);
        image.Dispose();
        var imageFrame = new ImageFrame(ImageFormat.Types.Format.Srgba, image.width, image.height, 4 * image.width, _buffer);
        var currentTimestamp = _stopwatch.ElapsedTicks / (TimeSpan.TicksPerMillisecond / 1000);
        var imageFramePacket = new ImageFramePacket(imageFrame, new Timestamp(currentTimestamp));
        _graph.AddPacketToInputStream("input_video", imageFramePacket).AssertOk();
        StartCoroutine(WaitForEndOfFrameCoroutine());
        var arlistX = new ArrayList(); // recommended 
        var arlistY = new ArrayList(); // recommended 
        if (_handLandmarkStream.TryGetNext(out var handLandmarks))
        {
            foreach (var landmarks in handLandmarks)
            {
                for (var i = 0; i < landmarks.Landmark.Count; i++)
                {
                    var worldLandmarkPos = _screenRect.GetPoint(landmarks.Landmark[i]);
                    hand[i].transform.localPosition = worldLandmarkPos;
                    arlistX.Add(worldLandmarkPos.x);
                    arlistY.Add(worldLandmarkPos.y);
                }
                generateObject(arlistX,arlistY);
            }
            
        }
    }
}

private IEnumerator WaitForEndOfFrameCoroutine()
{
    yield return new WaitForEndOfFrame();
}

private void generateObject(ArrayList alx,ArrayList aly){
    Vector2 down = new Vector2((float)alx[0],(float) aly[0]);
    Vector2 left = new Vector2((float)alx[4],(float)aly[4]); 
    Vector2 up = new Vector2((float)alx[12],(float) aly[12]); 
    Vector2 right = new Vector2((float)alx[20],(float) aly[20]); 
    Vector2 middle = new Vector2((float)alx[9],(float) aly[9]); 

    float distance1 = Vector2.Distance(down, middle);
    float distance2 = Vector2.Distance(left, middle);
    float distance3 = Vector2.Distance(up, middle);
    float distance4 = Vector2.Distance(right, middle);
    var distances = new ArrayList(); 
    distances.Add(distance1);
    distances.Add(distance2);
    distances.Add(distance3);
    distances.Add(distance4);
    float max = 0;
    for(var i=0;i<distances.Count;i++){
        if((float)distances[i] > max)
            max = (float)distances[i];
    }
    Vector3 position = new Vector3(middle.x, middle.y, 1.0f);
    Vector3 scale = new Vector3(max*1.5f,max*1.5f,0.1f);
    GameObject hand= GameObject.Find("HandObjectCollider");
    hand.transform.localPosition = position;
    hand.transform.localScale = scale;
}

private void alocBuffer(XRCpuImage image)
{
    var length = image.width * image.height * 4;
    if (_buffer == null || _buffer.Length != length)
    {
        _buffer = new NativeArray<byte>(length, Allocator.Persistent, NativeArrayOptions.UninitializedMemory);
    }
}

private void OnDestroy()
{
    _cameraManager.frameReceived -= OnCameraFrameReceived;
    var statusGraph = _graph.CloseAllPacketSources();
    if (!statusGraph.Ok())
    {
        Debug.Log($"Failed to close packet sources: {statusGraph}");
    }

    statusGraph = _graph.WaitUntilDone();
    if (!statusGraph.Ok())
    {
        Debug.Log(statusGraph);
    }

    _graph.Dispose();

    if (_gpuResources != null)
        _gpuResources.Dispose();

    _buffer.Dispose();
}

private void OnApplicationQuit()
{
    Protobuf.ResetLogHandler();
}

}`

Here is my scene:

Code to Reproduce the issue

No response

Additional Context

No response