How can I use ARCamera as the input of HandLandmarkDetection in the new version?

[price6856]

Feature Description

I have looked at the solutions of previous MediaPipe versions, but some classes or methods have been deprecated compared to the latest MediaPipe for Unity. Can you provide a simple example? Thank you.

Current Behaviour/State

No response

Additional Context

No response

[nemajor1]

Check this

void OnCameraFrameReceived(ARCameraFrameEventArgs args)
    {
        if (arCameraManager.TryAcquireLatestCpuImage(out XRCpuImage image))
        {
            var conversionParams = new XRCpuImage.ConversionParams
            {
                inputRect = new RectInt(0, 0, image.width, image.height),
                outputDimensions = new Vector2Int(image.width, image.height),
                outputFormat = TextureFormat.RGBA32,
                transformation = XRCpuImage.Transformation.None
            };

            var rawTextureData = new NativeArray<byte>(image.GetConvertedDataSize(conversionParams), Allocator.Temp);
            image.Convert(conversionParams, rawTextureData);

            if (cameraTexture == null || cameraTexture.width != image.width || cameraTexture.height != image.height)
            {
                cameraTexture = new Texture2D(image.width, image.height, TextureFormat.RGBA32, false);
            }

            cameraTexture.LoadRawTextureData(rawTextureData);
            cameraTexture.Apply();

            var textureFrame = new TextureFrame(cameraTexture.width, cameraTexture.height, TextureFormat.RGBA32);

            textureFrame.ReadTextureFromOnCPU(cameraTexture);

            graph.AddTextureFrameToInputStream(textureFrame);

            rawTextureData.Dispose();
            image.Dispose();
        }
    }
use XRCpuImage.ConversionParams and AddTextureFrameToInputStream, but there is a bug with coordinate inversion