Connecting to webcam using pygame module

NOTE: If you want to remove pygame dependency, you can go much lower level and access the Video 4 Linux 2 API directly using ioctl calls using Python's fcntl library.

In [1]:
import pygame
import pygame.camera as cm

pygame 1.9.4
Hello from the pygame community. https://www.pygame.org/contribute.html

1) Initialize pygame camera

In [2]:
cm.init()

2) Check if your device is listed

The device should have been listed before running the docker

docker run -it --device /dev/video0 ...

In case it gets complicated to find devices, use --privileged and map all devices

docker run -it --privileged -v /dev:/dev ...

To find the device folder in your machine (command line)

sudo apt-get install v4l-utils
v4l2-ctl --list-devices

OUTPUT : Webcam C170 (usb-0000:00:14.0-4): /dev/video0

You can also list devices from pygame module

In [3]:
cm.list_cameras()
Out[3]:
['/dev/video0']

3) Set up the device

In [4]:
cam = cm.Camera(cm.list_cameras()[0])

4) Start camera and capture pictures

In [5]:
cam.start()
In [6]:
img = cam.get_image()
In [7]:
img
Out[7]:
<Surface(640x480x24 SW)>
In [8]:
cam.stop()

5) Change the image to numpy array

Numpy should be installed separately

In [9]:
#Convert to numpy array
#The height and width are swapped
imgdata = pygame.surfarray.array3d(img)
In [10]:
imgdata
Out[10]:
array([[[155, 175, 175],
        [156, 173, 176],
        [161, 176, 180],
        ...,
        [ 56,  55,  34],
        [ 58,  53,  33],
        [ 58,  52,  37]],

       [[157, 177, 177],
        [159, 176, 179],
        [160, 175, 179],
        ...,
        [ 56,  55,  34],
        [ 59,  54,  34],
        [ 58,  52,  37]],

       [[160, 178, 176],
        [160, 176, 177],
        [162, 174, 181],
        ...,
        [ 55,  57,  35],
        [ 57,  56,  35],
        [ 55,  56,  39]],

       ...,

       [[ 27,  27,  27],
        [ 28,  27,  27],
        [ 29,  25,  26],
        ...,
        [ 21,  20,  17],
        [ 21,  20,  17],
        [ 21,  20,  17]],

       [[ 30,  26,  27],
        [ 30,  26,  24],
        [ 29,  25,  26],
        ...,
        [ 19,  19,  19],
        [ 20,  19,  19],
        [ 20,  19,  19]],

       [[ 29,  25,  26],
        [ 29,  25,  23],
        [ 29,  25,  26],
        ...,
        [ 19,  19,  19],
        [ 20,  19,  19],
        [ 20,  19,  19]]], dtype=uint8)
In [11]:
imgdata.shape
Out[11]:
(640, 480, 3)
In [12]:
import matplotlib.pyplot as plt
%matplotlib inline
In [13]:
#image dimensions are swapped
plt.imshow(imgdata)
Out[13]:
<matplotlib.image.AxesImage at 0x7fd696d42ba8>
In [14]:
plt.imshow(imgdata.swapaxes(0,1)) #Can also use transpose (1,0,2)
Out[14]:
<matplotlib.image.AxesImage at 0x7fd6963d89b0>