machidyo / mediapipehandtracking

This is based on the Java code and made to work with kotlin. And even more so, with MediaPipe, you can experience MediaPipe's multi-handed tracking.

mediapipehandtracking's Issues

Can I measure the size of the hand？

Is there a size formula for calculating hand size based on the data of 20 nodes

Getting hand landmarks without rendering data to the screen.?

Hi I am looking for output with the below points

I'm able to run the media pipe on android and get the landmarks, however, I'd like to see the output without the SurfaceView.
I just want the landmarks(in like a List<>) without any rendering on to the screen.

In 1st Point, if I disable frame layout rendering from the screen from XML I'm not able to deduct any hand points on

MediaPipeHandTracking/app/src/main/java/com/example/mediapipehandtracking/MainActivity.kt

Line 122 in 8fa599c

val viewGroup = findViewById<ViewGroup>(R.id.preview_display_layout)

How to edit camera screen

Good morning.
I would like to ask, how can I change layout of camera screen.
Activity_main.xml only shows screen of camera permission.
How to change camera layout.
Kindly help me.

How should I fix this mistake? E/tflite: Next operations are not supported by GPU delegate:

I/native: I0928 21:04:56.451962 25798 jni_util.cc:40] GetEnv: not attached
I/native: I0928 21:04:56.556967 25756 resource_util_android.cc:53] Successfully loaded: palm_detection.tflite
I/tflite: Initialized TensorFlow Lite runtime.
I/tflite: Created TensorFlow Lite delegate for GPU.
I/native: I0928 21:04:59.444133 25801 resource_util_android.cc:53] Successfully loaded: palm_detection_labelmap.txt
I/native: I0928 21:04:59.509568 25756 resource_util_android.cc:53] Successfully loaded: hand_landmark.tflite
E/tflite: Next operations are not supported by GPU delegate:
MEAN: Operation is not supported.
First 160 operations will run on the GPU, and the remaining 7 on the CPU.
E/tflite: external/org_tensorflow/tensorflow/lite/kernels/fully_connected.cc:105 filter->type != kTfLiteFloat32 (10 != 1)
Node number 155 (FULLY_CONNECTED) failed to prepare.
external/org_tensorflow/tensorflow/lite/kernels/conv.cc:272 bias->type != input_type (10 != 1)
Node number 2 (CONV_2D) failed to prepare.

How should I fix this mistake?