Face blendshapes MLP-Mixer (ONNX)
Google MediaPipe's face blendshape model (MLP-Mixer), converted to ONNX.
- Input
[1,146,2]— the 146-landmark subset as pixel coords (x·imgW, y·imgH). - Output
[52]— 52 ARKit-compatible blendshape scores in[0,1], in MediaPipe order (_neutral,browDownLeft, …jawOpen, …mouthSmileLeft/Right, …eyeBlinkLeft/Right, …). These drive ARKit-style morph targets directly.
Converted with tf2onnx optimizers disabled (the optimized graph broke a LayerNorm under ONNX Runtime — see the spike doc).
License
Apache-2.0 — this graph is a direct ONNX conversion of a Google
MediaPipe model (Apache-2.0 code
AND weights). Conversion + numerical-parity proof (vs the Python mediapipe
reference): scripts/export-facecap-onnx.py,
contract in docs/MOCAP_SPIKE.md.
How it is used
Mirror of one graph from fernandotonon/QtMeshEditor-models
(mocap/…), which QtMeshEditor
downloads on first use for its Performance Capture feature (video/webcam →
facial morph + head + full-body skeletal animation, epic #869). This standalone
repo is for discoverability; the app fetches from the aggregate repo.