# 安裝OpenCV

更新系統
$ sudo apt update && sudo apt upgrade -y

使用apt安裝
$ sudo apt install -y python3-opencv

### GPU記憶體更改



reboot後

### V4L2(Video4Linux2) and MMAL (Multi-Media Abstraction Layer)

在 Raspberry Pi OS Bullseye 之前，官方使用 raspivid 和 raspistill 來存取 Raspberry Pi Camera，這些工具基於 MMAL (Multi-Media Abstraction Layer) 和 V4L2 (Video4Linux2)。

但 從 Bullseye 版本開始（2021年後），Raspberry Pi 完全移除了 MMAL，轉而改用 libcamera 來管理相機。

### Picamera2（Python 版 libcamera）

官方為了讓 Python 開發者方便使用 libcamera，推出了 Picamera2，這是 Raspberry Pi 官方推薦的 Python API。

使用V4L2(現在已未使用)

cap = cv2.VideoCapture(0, cv2.CAP_V4L2)

使用libcamera

from picamera2 import Picamera2  

picam2 = Picamera2()
picam2.preview_configuration.main.size = (640, 480)  # Set resolution
picam2.preview_configuration.main.format = "RGB888"
picam2.configure("preview")
picam2.start()

while True:
    frame = picam2.capture_array()

## 下載 Haar Cascades（OpenCV 內建的人臉辨識模型）

wget -O haarcascade_frontalface_default.xml https://github.com/opencv/opencv/raw/master/data/haarcascades/haarcascade_frontalface_default.xml

此檔案須跟程式碼在同個資料夾(路徑)

### 輸出影像
影像格式未設定


### Code

:::spoiler face_detection.py

import cv2
from picamera2 import Picamera2
import numpy as np

# Load Haar Cascade for face detection
cascPath = "haarcascade_frontalface_default.xml"
faceCascade = cv2.CascadeClassifier(cascPath)

if faceCascade.empty():
    print("Error: Could not load cascade file. Check the file path.")
    exit()

# Initialize Picamera2
picam2 = Picamera2()
picam2.preview_configuration.main.size = (640, 480)  # Set resolution to 640x480
picam2.preview_configuration.main.format = "RGB888"  # Set image format to RGB888
picam2.configure("preview")  # Configure camera for preview mode
picam2.start()  # Start the camera

while True:
    frame = picam2.capture_array()  # Capture a frame from the camera
    if frame is None:
        print("Unable to read frame")
        break

    # Convert the captured frame from RGB to grayscale for face detection
    gray = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)

    # Detect faces in the grayscale image
    faces = faceCascade.detectMultiScale(
        gray,
        scaleFactor=1.1,  # Scale factor for face detection sensitivity
        minNeighbors=5,  # Minimum number of neighbor rectangles to retain a face detection
        minSize=(30, 30)  # Minimum size of detected faces
    )

    print("Found {0} faces!".format(len(faces)))  # Print number of faces detected

    # Draw rectangles around detected faces
    for (x, y, w, h) in faces:
        cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)

    # Display the frame with detected faces
    cv2.imshow("Face Detection", frame)

    # Press 'q' to exit the loop
    if cv2.waitKey(1) & 0xFF == ord("q"):
        break

# Stop the camera and close OpenCV windows
picam2.stop()
cv2.destroyAllWindows()

:::