How to Work with Camera Feed in Python

OpenCV makes it remarkably easy to start capturing video. The idea is simple: open the camera once, then loop forever, reading and displaying one frame at a time. That continuous loop is what turns a series of still images into live video.

import cv2

# Open the default webcam (index 0)
cap = cv2.VideoCapture(0)

while True:
    # Capture frame-by-frame
    ret, frame = cap.read()

    # Display the resulting frame
    cv2.imshow('Webcam Feed', frame)

    # Exit when 'q' is pressed
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# Release the capture
cap.release()
cv2.destroyAllWindows()

What each part does

Processing full-resolution video can be slow, especially once you start adding detection. Resizing each frame to a smaller size speeds everything up and helps the feed fit a fixed user interface. Add this line inside your loop, right after cap.read().

frame = cv2.resize(frame, (640, 480))  # width x height

Many computer vision tasks, including face and motion detection, work on grayscale images because colour is not needed and grayscale is faster to process. OpenCV converts between colour spaces with cvtColor.

gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
cv2.imshow('Grayscale Feed', gray)

Because each frame is just an image, you can draw shapes and text directly onto it before displaying. This is how detection boxes, labels and on-screen information get added to a live feed.

# Draw a green rectangle: top-left, bottom-right, colour (BGR), thickness
cv2.rectangle(frame, (50, 50), (200, 200), (0, 255, 0), 2)

# Draw red text: string, position, font, scale, colour (BGR), thickness
cv2.putText(frame, 'Live', (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 255), 2)

To record the feed to a file, OpenCV uses a VideoWriter. You tell it the codec, the output filename, the frame rate, and the frame size, then write each frame to it inside the loop.

import cv2

# Define the codec and create a VideoWriter object
fourcc = cv2.VideoWriter_fourcc(*'XVID')  # or 'MJPG'
out = cv2.VideoWriter('output.avi', fourcc, 20.0, (640, 480))

cap = cv2.VideoCapture(0)

while cap.isOpened():
    ret, frame = cap.read()
    if ret:
        out.write(frame)  # Save this frame to the file
        cv2.imshow('Recording...', frame)
        if cv2.waitKey(1) == ord('q'):
            break
    else:
        break

cap.release()
out.release()
cv2.destroyAllWindows()

Now for the exciting part. By combining the camera feed with a face detector, you can find and highlight faces in real time. This example uses dlib, a popular library with a fast and accurate frontal face detector.

import dlib

detector = dlib.get_frontal_face_detector()

while True:
    ret, frame = cap.read()
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

    faces = detector(gray)
    for face in faces:
        x, y, w, h = face.left(), face.top(), face.width(), face.height()
        cv2.rectangle(frame, (x, y), (x+w, y+h), (255, 0, 0), 2)

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

The detector runs on the grayscale frame for speed, returns a list of face regions, and then a blue rectangle is drawn around each one on the original colour frame.

The camera index you pass to VideoCapture selects which camera to use. Index 0 is the default built-in webcam. To use a second, external or USB camera, change the number.

cap = cv2.VideoCapture(1)  # or 2, 3, depending on your device

Find out how many cameras are connected

Not sure which index belongs to which camera? Loop through the first few indexes and check which ones respond.

for i in range(5):
    cap = cv2.VideoCapture(i)
    if cap.read()[0]:
        print(f"Camera {i} is available.")
    cap.release()

Live video trips up almost every beginner at some point. Here are the most common problems and how to fix them.

You now have every building block for working with a camera feed in Python. Here is the whole toolkit at a glance.

Project ideas to build next

The best way to lock in these skills is to build something. Here are five projects, each within reach now that you can capture and process live video.