Connexion
MediaPipe iris landmark detection model on Unity Barracuda. https://github.com/keijiro/IrisBarracuda
MediaPipe iris landmark detection model on Unity Barracuda. https://github.com/keijiro/IrisBarracuda
Iris landmark detection model running on Unity Barracuda https://github.com/keijiro/IrisBarracuda
Iris landmark detection model running on Unity Barracuda https://github.com/keijiro/IrisBarracuda
MediaPipe iris landmark detection model on Unity Barracuda https://github.com/keijiro/IrisBarracuda
Added head rotation support (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
Added head rotation support (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
BlazeFace + Face Mesh on Unity Barracuda https://github.com/keijiro/FaceMeshBarracuda
BlazeFace + Face Mesh on Unity Barracuda https://github.com/keijiro/FaceMeshBarracuda
The Face Mesh model is impressively fast. It runs smoothly with 60 fps video input. (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
The Face Mesh model is impressively fast. It runs smoothly with 60 fps video input. (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
Implemented face region cropping. Now the pipeline works with real time video input. (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
Implemented face region cropping. Now the pipeline works with real time video input. (BlazeFace + Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
Face Mesh triangulation (MediaPipe Face Mesh on Unity Barracuda) https://github.com/keijiro/FaceMeshBarracuda
Wrote a Colab notebook to convert the MediaPipe Face Mesh model. https://colab.research.google.com/drive/1C6zEB3__gcHEWnWRm-b4jIA0srA1gkyq?usp=sharing
Finally, the MediaPipe Face Mesh model worked on Unity Barracuda.
Implemented weighted NMS for BlazeFace on Unity Barracuda. https://github.com/keijiro/BlazeFaceBarracuda
Implemented weighted NMS for BlazeFace on Unity Barracuda. https://github.com/keijiro/BlazeFaceBarracuda
Looking into the MediaPipe hair segmentation model. I think I have to replace this ScatterND node to add Barracuda support.
So now I can make face filters (BlazeFace on Unity Barracuda) https://github.com/keijiro/BlazeFaceBarracuda
So now I can make face filters (BlazeFace on Unity Barracuda) https://github.com/keijiro/BlazeFaceBarracuda
Added key point tracking to the Barracuda BlazeFace implementation. I haven’t implemented the weighted suppression yet, but it already looks good enough to me. https://github.com/keijiro/BlazeFaceBarracuda
Added key point tracking to the Barracuda BlazeFace implementation. I haven’t implemented the weighted suppression yet, but it already looks good enough to me. https://github.com/keijiro/BlazeFaceBarracuda
Converted the BlazeFace model. It seems to return correct inference results now. (left: Barracuda on Unity, right: onnxruntime on Colab)
Implementing an ONNX converter to add Barracuda support to the BlazeFace model. This is the first time I use the Python ONNX module.
Trying dynamic virtual backgrounds with the MediaPipe Meet segmentation model (running on Unity Barracuda). https://github.com/keijiro/MeetBarracuda
Trying dynamic virtual backgrounds with the MediaPipe Meet segmentation model (running on Unity Barracuda). https://github.com/keijiro/MeetBarracuda
Collanews 