3ddfa

Towards Fast, Accurate and Stable 3D Dense Face Alignment

Input

(Image from https://github.com/cleardusk/3DDFA_V2/blob/master/examples/inputs/emma.jpg)

Output

Setup

Build the the faster mesh render. (for 3d drawing)

sh ./build.sh

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 3ddfa.py

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 3ddfa.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 3ddfa.py --video VIDEO_PATH

By adding the --mode option, you can specify output option which is selected from "2d_sparse", "2d_dense", "3d", "pose". (default is 2d_sparse)

$ python3 3ddfa.py --mode 2d_sparse

Reference

• Towards Fast, Accurate and Stable 3D Dense Face Alignment

Framework

Pytorch

Model Format

ONNX opset=14

Netron

mb1_120x120.onnx.prototxt
FaceBoxesProd.onnx.prototxt
bfm_noneck_v3.onnx.prototxt