【你购物我买单】在Arduino UNO Q上用MediaPipe实现面部识别

概述

本教程详细记录了在arduino uno q上安装Miniconda、配置Python环境、安装MediaPipe库，并编写一个简单的面部识别程序的全过程。特别针对ARM架构的兼容性问题提供了解决方案。

硬件

硬件：arduino uno q

步骤一：安装Miniconda

1.1 下载正确的Miniconda版本

由于arduino uno q 是ARM架构，不能使用标准的x86_64版本。

确认是aarch64架构后，下载ARM版本的Miniconda：

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-aarch64.sh

1.2 安装过程

运行安装脚本，按照提示完成安装：

bash Miniconda3-latest-Linux-aarch64.sh

安装注意事项：

接受许可协议

使用默认安装路径 /home/arduino/miniconda3

选择"yes"初始化conda

1.3 初始化conda

安装完成后，需要重新加载shell配置或重启终端：

source /home/arduino/miniconda3/bin/activate base

验证安装：

conda --version

步骤二：配置Python环境

2.1 创建Python 3.9环境

为了避免与系统Python冲突，创建独立的conda环境：

conda create -n py39 python=3.9 
conda activate py39

2.2 安装必要的依赖

pip install numpy opencv-python

步骤三：安装MediaPipe

3.1 安装MediaPipe

由于MediaPipe官方支持ARM架构，可以直接通过pip安装：

pip install mediapipe

步骤四：编写面部识别程序

4.1 基础面部检测程序

创建 face_detection.py 文件：

python

import cv2
import mediapipe as mp

def check_detection_details(image_path):
    mp_face_detection = mp.solutions.face_detection
    mp_drawing = mp.solutions.drawing_utils
    
    image = cv2.imread(image_path)
    if image is None:
        print("Cannot read image")
        return
    
    image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    
    # Lower confidence threshold to increase detection sensitivity
    with mp_face_detection.FaceDetection(
        model_selection=0,  # Try short-range detection model
        min_detection_confidence=0.3) as face_detection:  # Lower threshold
        
        results = face_detection.process(image_rgb)
        
        print(f"Detection results: {results.detections}")
        if results.detections:
            print(f"Detected {len(results.detections)} faces")
            for i, detection in enumerate(results.detections):
                confidence = detection.score[0]
                print(f"Face {i+1}: Confidence = {confidence:.4f}")
                
                # Draw detection results
                mp_drawing.draw_detection(image, detection)
                
                # Manually draw a more obvious bounding box
                bbox = detection.location_data.relative_bounding_box
                h, w, _ = image.shape
                x = int(bbox.xmin * w)
                y = int(bbox.ymin * h)
                width = int(bbox.width * w)
                height = int(bbox.height * h)
                
                # Draw a thick red rectangle
                cv2.rectangle(image, (x, y), (x + width, y + height), (0, 0, 255), 3)
                cv2.putText(image, f"Face {i+1}: {confidence:.2f}", 
                           (x, y-10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 2)
        else:
            print("No faces detected!")
            # For debugging, show the image dimensions
            print(f"Image dimensions: {image.shape}")
            
            # Try to see if image is being processed correctly
            # Save the RGB image to check
            test_output = image_path.replace('.jpg?x-oss-process=image/watermark,g_center,image_YXJ0aWNsZS9wdWJsaWMvd2F0ZXJtYXJrLnBuZz94LW9zcy1wcm9jZXNzPWltYWdlL3Jlc2l6ZSxQXzQwCg,t_20', '_test_rgb.jpg?x-oss-process=image/watermark,g_center,image_YXJ0aWNsZS9wdWJsaWMvd2F0ZXJtYXJrLnBuZz94LW9zcy1wcm9jZXNzPWltYWdlL3Jlc2l6ZSxQXzQwCg,t_20')
            cv2.imwrite(test_output, image_rgb)
            print(f"RGB test image saved to: {test_output}")
    
    # Save the result
    output_path = image_path.replace('.jpg?x-oss-process=image/watermark,g_center,image_YXJ0aWNsZS9wdWJsaWMvd2F0ZXJtYXJrLnBuZz94LW9zcy1wcm9jZXNzPWltYWdlL3Jlc2l6ZSxQXzQwCg,t_20', '_debug.jpg?x-oss-process=image/watermark,g_center,image_YXJ0aWNsZS9wdWJsaWMvd2F0ZXJtYXJrLnBuZz94LW9zcy1wcm9jZXNzPWltYWdlL3Jlc2l6ZSxQXzQwCg,t_20')
    cv2.imwrite(output_path, image)
    print(f"Debug result saved to: {output_path}")
    
    # Try to display
    try:
        cv2.imshow('Debug Detection', image)
        cv2.waitKey(0)
        cv2.destroyAllWindows()
    except:
        print("Display not available")

# Run the debug function
check_detection_details("/home/arduino/xxg.jpg?x-oss-process=image/watermark,g_center,image_YXJ0aWNsZS9wdWJsaWMvd2F0ZXJtYXJrLnBuZz94LW9zcy1wcm9jZXNzPWltYWdlL3Jlc2l6ZSxQXzQwCg,t_20")