**Note**: This is a tracking library, **not** a stand-alone avatar puppeteering program. I'm also working on [VSeeFace](https://www.vseeface.icu/), which allows animating [VRM](https://vrm.dev/en/how_to_make_vrm/) and [VSFAvatar](https://www.youtube.com/watch?v=jhQ8DF87I5I) 3D models by using OpenSeeFace tracking. [VTube Studio](https://denchisoft.com/) uses OpenSeeFace for webcam based tracking to animate Live2D models.
As Pytorch 1.3 CPU inference speed on Windows is very low, the model was converted to ONNX format. Using [onnxruntime](https://github.com/microsoft/onnxruntime) it can run at 30 - 60 fps tracking a single face. There are four models, with different speed to tracking quality trade-offs.
An up to date sample video can be found [here](https://www.youtube.com/watch?v=AaNap_ud_3I&vq=hd1080), showing the default tracking model's performance under different noise and light levels.
The face tracking itself is done by the `facetracker.py` Python 3.7 script. It is a commandline program, so you should start it manually from cmd or write a batch file to start it. If you downloaded a release and are on Windows, you can run the `facetracker.exe` inside the `Binary` folder without having Python installed. You can also use the `run.bat` inside the `Binary` folder for a basic demonstration of the tracker.
The script will perform the tracking on webcam input or video file and send the tracking data over UDP. This design also allows tracking to be done on a separate PC from the one who uses the tracking information. This can be useful to enhance performance and to avoid accidentially revealing camera footage.
The provided `OpenSee` Unity component can receive these UDP packets and provides the received information through a public field called `trackingData`. The `OpenSeeShowPoints` component can visualize the landmark points of a detected face. It also serves as an example. Please look at it to see how to properly make use of the `OpenSee` component. Further examples are included in the `Examples` folder. The UDP packets are received in a separate thread, so any components using the `trackingData` field of the `OpenSee` component should first copy the field and access this copy, because otherwise the information may get overwritten during processing. This design also means that the field will keep updating, even if the `OpenSee` component is disabled.
Run the python script with `--help` to learn about the possible options you can set.
python facetracker.py --help
A simple demonstration can be achieved by creating a new scene in Unity, adding an empty game object and both the `OpenSee` and `OpenSeeShowPoints` components to it. While the scene is playing, run the face tracker on a video file:
The included `OpenSeeLauncher` component allows starting the face tracker program from Unity. It is designed to work with the pyinstaller created executable distributed in the binary release bundles. It provides three public API functions:
*`public string[] ListCameras()` returns the names of available cameras. The index of the camera in the array corresponds to its ID for the `cameraIndex` field. Setting the `cameraIndex` to `-1` will disable webcam capturing.
*`public bool StartTracker()` will start the tracker. If it is already running, it will shut down the running instance and start a new one with the current settings.
*`public void StopTracker()` will stop the tracker. The tracker is stopped automatically when the application is terminated or the `OpenSeeLauncher` object is destroyed.
The `OpenSeeLauncher` component uses WinAPI job objects to ensure that the tracker child process is terminated if the application crashes or closes without terminating the tracker process first.
Additional custom commandline arguments should be added one by one into elements of `commandlineArguments` array. For example `-v 1` should be added as two elements, one element containing `-v` and one containing `1`, not a single one containing both parts.
The `OpenSeeExpression` component can be added to the same component as the `OpenSeeFace` component to detect specific facial expressions. It has to be calibrated on a per-user basis. It can be controlled either through the checkboxes in the Unity Editor or through the equivalent public methods that can be found in its source code.
To calibrate this system, you have to gather example data for each expression. If the capture process is going too fast, you can use the `recordingSkip` option to slow it down.
To save both the trained model and the captured training data, type in a filename including its full path in the "Filename" field and tick the "Save" box. To load it, enter the filename and tick the "Load" box.
* A reasonable number of expressions is six, including the neutral one.
* Before starting to capture expressions, make some faces and wiggle your eyebrows around, to warm up the feature detection part of the tracker.
* Once you have a detection model that works decently, when using it take a moment to check all the expressions work as intended and add a little data if not.
* The highest quality model is selected with `--model 3`, the fastest model with the lowest tracking quality is `--model 0`.
* Lower tracking quality mainly means more rigid tracking, making it harder to detect blinking and eyebrow motion.
* Depending on the frame rate, face tracking can easily use up a whole CPU core. At 30fps for a single face, it should still use less than 100% of one core on a decent CPU. If tracking uses too much CPU, try lowering the frame rate. A frame rate of 20 is probably fine and anything above 30 should rarely be necessary.
* When setting the number of faces to track to a higher number than the number of faces actually in view, the face detection model will run every `--scan-every` frames. This can slow things down, so try to set `--faces` no higher than the actual number of faces you are tracking.
Four pretrained face landmark models are included. Using the `--model` switch, it is possible to select them for tracking. The given fps values are for running the model on a single face video on a single CPU core. Lowering the frame rate would reduce CPU usage by a corresponding degree.
More samples: [Results3.png](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/Results3.png), [Results4.png](https://raw.githubusercontent.com/emilianavt/OpenSeeFace/master/Images/Results4.png)
## Face detection
The landmark model is quite robust with respect to the size and orientation of the faces, so the custom face detection model gets away with rougher bounding boxes than other approaches. It has a favorable speed to accuracy ratio for the purposes of this project.
The builds in the release section of this repository contain a `facetracker.exe` inside a `Binary` folder that was built using `pyinstaller` and contains all required dependencies.
To run it, at least the `models` folder has to be placed in the same folder as `facetracker.exe`. Placing it in a common parent folder should work too.
When distributing it, you should also distribute the `Licenses` folder along with it to make sure you conform to requirements set forth by some of the third party libraries. Unused models can be removed from redistributed packages without issue.
The model was trained on a 66 point version of the [LS3D-W](https://www.adrianbulat.com/face-alignment) dataset.
@inproceedings{bulat2017far,
title={How far are we from solving the 2D \& 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)},
author={Bulat, Adrian and Tzimiropoulos, Georgios},
booktitle={International Conference on Computer Vision},
year={2017}
}
Additional training has been done on the WFLW dataset after reducing it to 66 points and replacing the contour points and tip of the nose with points predicted by the model trained up to this point. This additional training is done to improve fitting to eyes and eyebrows.
@inproceedings{wayne2018lab,
author = {Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
title = {Look at Boundary: A Boundary-Aware Face Alignment Algorithm},
For the training the gaze and blink detection model, the [MPIIGaze](https://www.mpi-inf.mpg.de/departments/computer-vision-and-machine-learning/research/gaze-based-human-computer-interaction/appearance-based-gaze-estimation-in-the-wild/) dataset was used. Additionally, around 125000 synthetic eyes generated with [UnityEyes](https://www.cl.cam.ac.uk/research/rainbow/projects/unityeyes/) were used during training.
* [Designing Neural Network Architectures for Different Applications: From Facial Landmark Tracking to Lane Departure Warning System](https://www.synopsys.com/designware-ip/technical-bulletin/ulsee-designing-neural-network.html) by YiTa Wu, Vice President of Engineering, ULSee
* [Real-time Human Pose Estimation in the Browser with TensorFlow.js](https://blog.tensorflow.org/2018/05/real-time-human-pose-estimation-in.html)
* [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://lmb.informatik.uni-freiburg.de/people/ronneber/u-net/) by Olaf Ronneberger, Philipp Fischer, Thomas Brox
* [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861) by Andrew G. Howard, Menglong Zhu, Bo Chen, Dmitry Kalenichenko, Weijun Wang, Tobias Weyand, Marco Andreetto, Hartwig Adam
* [Searching for MobileNetV3](https://arxiv.org/abs/1905.02244) by Andrew Howard, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang, Yukun Zhu, Ruoming Pang, Vijay Vasudevan, Quoc V. Le, Hartwig Adam
RetinaFace detection is based on [this](https://github.com/biubug6/Pytorch_Retinaface) implementation. The pretrained model was modified to remove unnecessary landmark detection and converted to ONNX format for a resolution of 640x640.
