liske / virtual_webcam_background

Use a virtual webcam background with body-pix and v4l2loopback

Virtual Webcam Background

Virtual Webcam Background allows you to use a virtual background or to blur the background of your webcam image similar to commercial programs like Zoom.

Tensorflow with BodyPix is used to segment the image into foreground (person) and background using a neural network and v4l2loopback is used to create a virtual webcam.

As the script creates a virtual webcam device it works with any program that can use a v4l2 webcam.

Installation

Install the requirements:

pip install -r requirements.txt

Due to packaging issues, you need to install tfjs-to-tf using

pip install tensorflowjs
pip install "git+https://github.com/patlevin/tfjs-to-tf.git@v0.5.0"

Download the bodypix model:

./get-model.sh

Then install v4l2loopback and load the kernel module:

modprobe v4l2loopback exclusive_caps=1 video_nr=2 # creates /dev/video2

The exclusive_caps option is needed by some programs, such as chromium.

Then copy config.yaml.example to config.yaml and edit the config as needed and run the virtual_webcam.py script.

If you have a Nvidia graphics card, you may want to install CUDA for better performance.

Configuration

To configure the virtual webcam, edit config.yaml. Most options are applied instantly, except for width and height as the webcam must be reinitialized to change them.

• width: The input resolution width.
• height: The input resolution height.
• segmentation_threshold: The threshold for foreground / background segmentation.
• blur: Blur factor for the mask to smooth the edges.
• dilate: Number of pixels the mask is shrunk to remove spots.
• erode: Number of pixels the mask is grown after shrinking to capture the full body image again.
• background_filters: Filters applied to the background or virtual background.
• foreground_filters: Filters applied to the detected foreground.
• result_filters: Filters applied to the result image.
• background_image: Filename of an image file or a directory containing images for an animation.
• overlay_image: Filename of an image file with alpha channel (transparency) or a directory containing images for an animation.
• background_fps: Maximal framerate for the animation of the background image.
• overlay_fps: Maximal framerate for the animation of the overlay image.
• virtual_video_device: The virtual video device, e.g., /dev/video2.
• real_video_device: The video device of your webcam, e.g. /dev/video0.
• average_masks: Number of masks to average. A higher number will result in afterimages, a smaller number in flickering at the boundary between foreground and background.
• flip_horizontal: Flip the input image horizontally.
• flip_vertical: Flip the input image vertically.
• background_interpolation_method: Interpolation method to use. Currently supported methods are BILINEAR and NEAREST. Usually BILINEAR is an good option, but for using pixel art backgrounds NEAREST may look better.
• debug_show_mask: Debug option to show the mask, that can be used to configure blur/dilate/erode correctly.

Note: Input width and height are autodetected when they are not set in the config, but this can lead to bad default values, e.g. 640x480 even when the camera supports a resolution of 1280x720.

Filters

The background_filters option is a list of filters that will be applied after each other.

A simple example that converts the background to grayscale and blurs it:

- `background_filters = ["grayscale", "blur"]`

Some filters have arguments. To change the blur value in the filter list above, use

- `background_filters = ["grayscale", ["blur", 10, 10]]`

Alternative syntax variants:

- `background_filters = ["grayscale", ["blur", [10, 10]]]`
- `background_filters = ["grayscale", ["blur", {intensity_x: 10, intensity_y: 10}]]`

Filters

The current filters and their options are:

• blur: Blur the image.
  • intensity_x: The intensity in x direction.
  • intensity_y: The intensity in y direction. When only intensity_x is given, it will be used for intensity_y as well.
• grayscale: Convert the image into a grayscale image.

Animations

Animations are a sequence of images in a folder. The script simply trys to load image_name/*.* when image_name is a folder. You need to make sure, that the folder only contains images and that the images are ordered correctly when they are ordered by their filename.

Example for creating an animation from a short video:

mkdir animation
cd animation
ffmpeg -i ../animation.gif -vf fps=10 out%04d.png

Animations currently run at the framerate of the virtual webcam, so their speed depends on the framerate that your computer is able to achieve for the virtual webcam. When using the ffmpeg command, you can adapt the framerate of the video to the framerate of the virtual webcam using the fps parameter.

Note that the script loads all images of an animation into RAM scaled to the resolution of your webcam, so using too long animations is not a good idea.

Advanced

You can try alternative models by editing the source code.

To download other models get the full get-model.sh script from https://github.com/ajaichemmanam/simple_bodypix_python and run it with one of these combinations:

./get-model.sh bodypix/mobilenet/float/{025,050,075,100}/model-stride{8,16}

Then edit the script and change output_stride and internal_resolution accordingly.

You can also try the resnet50 models, but then you will in addition need to change the preprocessing. The needed preprocessing for resnet50 is included as a comment in the source code.

Acknowledgements

• The program is inspired by this blog post by Benjamin Elder.
• Linux-Fake-Background-Webcam is a direct implementation of the blog post using docker and nodejs in addition to python.
• simple_bodypix_python was the base for an earlier version of the script.
• The functions in bodypix_functions.py are adapted from the body-pix nodejs module.