Moskari/pykinect2

kinect2 python wrapper

PyKinect2

python wrapper for kinect2 allow easier access of kinect2

Install from source

Install dependencies from libfreenect2

sudo apt-get install build-essential cmake pkg-config libturbojpeg libjpeg-turbo8-dev mesa-common-dev freeglut3-dev libxrandr-dev libxi-dev python3-dev swig
sudo apt-get install libturbojpeg0-dev
sudo apt-get install libusb-1.0-0-dev

Copy 90-kinect2.rules under /etc/udev/rules.d/ to grant permission for Kinect2 usb

Running pip should be enough:

pip install .

Examples

examples on how to fetch rgb images and depth map from kinect2 and the registered frame

from kinect2 import Kinect2
import cv2
import numpy as np


def main():
  kinect = Kinect2()
  if not kinect.IsOpened():  # check if kinect2 device is connected
    print('no kinect2 device found')
    return

  while True:
    kinect.NextFrame()  # grab the data from libfreenect2 frame
    rgb = kinect.RGB()
    ir = kinect.IR()
    depth = kinect.Depth()
    undistorted = kinect.Undistorted()
    registered = kinect.Registered()

    rgb = np.reshape(np.ravel(rgb, 'F'), rgb.shape)
    ir = np.reshape(np.ravel(ir, 'F'), ir.shape) / 1000.0
    depth = np.reshape(np.ravel(depth, 'F'), depth.shape)
    undistorted = np.reshape(np.ravel(undistorted, 'F'), undistorted.shape)
    registered = np.reshape(np.ravel(registered, 'F'), registered.shape)

    cv2.imshow('RGB', rgb)
    cv2.imshow('IR', ir)
    cv2.imshow('Depth', depth)
    cv2.imshow('Undistorted', undistorted)
    cv2.imshow('Registered', registered)
    key = cv2.waitKey(30)
    if key in [ord('q'), ord('Q')]:
      break
  kinect.Stop()


if __name__ == '__main__':
  main()