pyfakewebcam

An API for writing RGB frames to a fake webcam device on Linux!

Compatible with Python2.7 and Python3.x

Author: John Emmons
Email: mail@johnemmons.com

installation

# use pip to get the latest stable release
pip install pyfakewebcam

# use git to install the latest version
git clone https://github.com/jremmons/pyfakewebcam.git
cd pyfakewebcam
python setup.py install

dependencies

# python 
pip install numpy

# linux
apt-get install v4l2loopback-utils

# linux (optional)
apt-get install python-opencv # 10x performance improvement if installed (see below)
apt-get install ffmpeg # useful for debugging

performance

When I run the examples/example.py script (640x360 resolution) on an Intel i7-3520M (2.9GHz, turbos to 3.6 GHz), the time to schedule a single frame is ~3 milliseconds (with opencv installed). You can use this library without installing opencv, but it is almost 10x slower; time to schedule a frame without opencv is ~26 milliseconds (RGB to YUV conversion done with numpy operations).

If your goal is to run at 30Hz (or slower) and ~26 milliseconds of delay is acceptable in your application, then opencv may not be necessary.

usage

Insert the v4l2loopback kernel module.

modprobe v4l2loopback devices=2 # will create two fake webcam devices

Example code.

# see red_blue.py in the examples dir
import time
import pyfakewebcam
import numpy as np

blue = np.zeros((480,640,3), dtype=np.uint8)
blue[:,:,2] = 255

red = np.zeros((480,640,3), dtype=np.uint8)
red[:,:,0] = 255

camera = pyfakewebcam.FakeWebcam('/dev/video1', 640, 480)

while True:

    camera.schedule_frame(red)
    time.sleep(1/30.0)

    camera.schedule_frame(blue)
    time.sleep(1/30.0)

Run the following command to see the output of the fake webcam.

ffplay /dev/video1