How to get live video from your webcam?

Howard Pong Yuen Lam · AI Notes · November 1, 2026 · 4 min read

English 繁體中文 简体中文

The camera on your computer is the starting point for anything that has to look at the real world and then act on it. The lipstick example shows how: a video is many still frames arriving one after another. Find the face in each frame, put lipstick on it, and show that frame before moving on to the next. Do that for every frame, and the result is a moving face with lipstick on it.

The program below is the foundation: 10 lines that open your webcam and show the live video in a window, until you press q or Q. It contains no AI. It only gets frames from the camera and shows them, deliberately, so that the camera part is clear before AI is added on top.

What your eye sees and what the program sees

Your eye sees a smooth moving image. The program sees a long series of separate frames, each replacing the one before so quickly that the change is invisible.

How the program works

1. Turn on the camera

import cv2
cap = cv2.VideoCapture(0)

cv2 is OpenCV (Open Source Computer Vision), the library used throughout this series to open photos and video. The first line loads it.

The second line connects to a camera and keeps that connection in cap, short for capture. The 0 means the first camera on the computer, which is usually the built-in webcam. On a Mac, the computer asks for permission the first time you run it; on Windows, check that camera access is allowed in Settings if nothing appears.

2. Read one frame at a time

while True:
    ret, img = cap.read()
    cv2.imshow("Frame", img)

The first line starts a loop that never stops by itself, because True is always true. Everything indented under it runs again and again, once per frame.

cap.read() takes one frame. ret (return) says whether a frame was read, and img (image) is the frame itself: a grid of pixel values. That img is the frame you would process in any camera-based program — exactly like the photo files in the earlier notes.

cv2.imshow shows the frame in a window titled "Frame". Each new frame replaces the previous one, so the window looks like live video.

3. Stop when you press q

key = cv2.waitKey(1)
if key in (ord('q'), ord('Q')):
    break

cv2.waitKey(1) waits 1 millisecond for a key press, then continues. key is a code that tells you which key was pressed, or -1 if no key was pressed. This short wait also gives the window time to draw the frame, which is why the video appears at all.

The next line checks for q or Q. ord() turns a letter into its number — ord('q') is 113 — so a letter can be compared with the number in key. Testing both means Caps Lock makes no difference.

break leaves the loop and moves on to the last two lines. After you run the program, the terminal in Visual Studio Code is still the selected window, so a q or Q you type goes to Visual Studio Code, not to the program that shows the video. Move your mouse pointer over the video window and click once to select it. Then type q or Q to end the program.

There is also a quicker way to stop a running program: in the Visual Studio Code terminal, press the two keys Ctrl and C at the same time.

4. Turn the camera off

cap.release()
cv2.destroyAllWindows()

cap.release() disconnects from the camera, so other programs can use it again and the camera light goes out. cv2.destroyAllWindows() closes the window.

The whole program

The pieces above, put back together. This is the complete file, 10 lines in all:

import cv2
cap = cv2.VideoCapture(0)
while True:
    ret, img = cap.read()
    cv2.imshow("Frame", img)
    key = cv2.waitKey(1)
    if key in (ord('q'), ord('Q')):
        break
cap.release()
cv2.destroyAllWindows()

Save one frame

Many jobs need a single frame, not a stream: take one frame, then analyse it. You do not need a different program for that. Inside the loop, img is already one frame, so one extra line saves it to a file:

cv2.imwrite("camera_frame.jpg", img)

Add that line just after cap.read(), and the program writes a new file for every frame. To keep just one, add it above the break line instead, so the file is saved when you press q: what you see on screen at that moment is what you get.

Once the frame is a file, everything in the earlier notes applies to it — including face_recognition.load_image_file in How does AI find the key points on your face?.

Running it on your own computer

Download the zip file and unzip it to a folder. The zip holds all six programs in this series; this note uses face05.py. This one does not need a photo in the folder. Open the folder in Visual Studio Code, then type these three lines in the terminal:

uv python pin 3.12.10
uv sync
uv run face05.py

The first line records which Python version this folder uses, and the second installs every package that pyproject.toml lists, including OpenCV. The third line opens the camera window. Press q in that window to close it.

If you have not set up uv yet, start with How to install Visual Studio Code? and then the uv note for Windows or Mac.

What comes next

The camera is only the eye; the thinking comes next, in the note that adds the AI on top.