Accessing Camera

This page will explain how to access the camera in your computer.

Setup

Once again before we start activate your venv. Even though in this page we will not be using the venv remember to always activate it if you believe it is important.

I will activate it because I will be using jupyter notebook to document this process and I will use the venv kernel from within the jupyter notebook

~AI\computer_vision\od_projects\opencv_learn>venv_opencv\Scripts\activate

(venv_opencv) ~AI\computer_vision\od_projects\opencv_learn>jupyter notebook

.

Script

  • We start off by importing Opencv and
  • The systems module
  • specify camera device of 0
  • Then we check if there is any other command that is already in place that will override that value
  • Next we call the VideoCapture() class to capture a video object and name if source and we pass the device index into that class (s)
  • A default value of 0 will access the default video camera on out system, if you had more than one, then you’d direct it to whichever device you want 1 being the second camera, 2 being the third….
  • We then create a named window where we will send the screened output to
  • The while loop will continually send the signal to our output window until the user hits the ESC key: ==27 is the ESC key
  • The first line in loop uses the video capture class: source to call the video capture method: read to capture the first frame and return it as well as a logical var : has_frame
  • If the var logical returns a false: has_frame=false it will break from the loop otherwise it will continue on
  • calling the imshow() to send the output frame to the window
  • This works fine from CMd but doesn’t work from within jupyter notebook - The issue seems to be with the if len() code chunk. I’ll comment it and set the default value to 0. This is just for jupyter notebook only, it seems to work just find in CMD as is
import cv2
import sys

s = 0
#if len(sys.argv) > 1:    -- commented out for jupyter notebook only, it works with the if stmt in cmd
#    s = sys.argv[1]

source = cv2.VideoCapture(s)

win_name = 'YashYa Feed'
cv2.namedWindow(win_name, cv2.WINDOW_NORMAL)

while cv2.waitKey(1) != 27: # Escape
    has_frame, frame = source.read()
    if not has_frame:
        break
    cv2.imshow(win_name, frame)

source.release()
cv2.destroyWindow(win_name)

Another Script

This works in jupyter notebook. Use q to exit

import cv2

cap = cv2.VideoCapture(0)
if not cap.isOpened():
    print("Cannot open camera")
    exit()

while True:
    ret, frame = cap.read()
    if not ret:
        print("Can't receive frame (stream end?). Exiting ...")
        break

    cv2.imshow('Frame', frame)
    if cv2.waitKey(1) == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()

Separate Frames

  • If for some reason you need to plot out each frame in a separate output use matplotlib to plot each frame separately
# This shows frame by frame in a different plot, so each frame will be plotted separately
# If you only want one frame take the plt lines out of the loop
import cv2
import sys
from matplotlib import pyplot as plt
source = cv2.VideoCapture(0)

while cv2.waitKey(1) != 27: # Escape
    has_frame, frame = source.read()
    if not has_frame:
        break
    plt.imshow(frame)
    plt.title('YashaYa Feed')
    plt.show()

source.release()