Babylon with mediapipe

I used Babylon.js+kalidokit+mediapipe to implement a vtuber, but I found that my demo seems to be unable to accurately model the movement. Kalidokit said that it is no longer maintained. It is recommended to use mediapipe. Is there any reference related to mediapipe/holistic+Babylon.js? ?
I refer to kaolidokit for the part of the mediapipe driving the model, but I find that I can’t seem to drive the fingers of the model, and the movement points of the model are not very accurate. At present, it seems that GitHub - ButzYung/SystemAnimatorOnline: XR Animator, AI-based Full Body Motion Capture and Extended Reality (XR) solution, powered by System Animator Online is based on the three.js driver model and has the most accurate points. However, since this warehouse is not completely open source, it is impossible to refer to the point algorithm logic.

kalidokit: GitHub - yeemachine/kalidokit: Blendshape and kinematics calculator for Mediapipe/Tensorflow.js Face, Eyes, Pose, and Finger tracking models.

Here is my repository: GitHub - Spencer17x/babylon-vtuber

There’s a few threads about mediapipe in the forum, maybe they could help? For eg:

and others.