Note: This repository has moved to https://github.com/ORB-HD/deface. The copy you are viewing right now will no longer be updated and will be removed in the future.
From PyPI (official releases):
$ pip install deface
Current git master:
$ pip install 'git+https://github.com/mdraw/deface'
Basic usage after installing the deface
package:
$ deface <input.mp4> -o <output.mp4>
Show help and all options:
$ deface -h
Hardware acceleration with ONNX Runtime
If you have a CUDA-capable GPU, you can enable GPU acceleration by installing the relevant packages:
$ pip install onnx onnxruntime-gpu
If the onnxruntime-gpu
package is found and a GPU is available, the face detection network is automatically offloaded to the GPU.
This can significantly improve the overall processing speed.
If your machine doesn't have a CUDA-capable GPU but you want to accelerate computation on another hardware platform (e.g. Intel CPUs), you can look into the available options in the ONNX Runtime build matrix.
centerface.onnx
(original) andcenterface.py
(modified) are based on https://github.com/Star-Clouds/centerface (revision 8c39a49), released under MIT license