查看: 15217|回复: 5

[教程] 树莓派+mediapipe+pid 摄像头人脸跟随

本帖最后由云天于 2022-2-5 12:47 编辑

树莓派+mediapipe+pid 摄像头人脸跟随图4

本教程中使用了“树莓派4b”+“树莓派两自由度云台”+“mediapipe”+“pid”，实现摄像头人脸跟随。
【树莓派】
4代B型4GB Raspberry Pi

树莓派最新发布的第四代产品 Raspberry Pi 4 B，性能与树莓派 3B+相比无论是处理器速度，还是多媒体和内存上都有显著提升。Raspberry Pi 4 B拥有与入门级 x86 PC 系统相媲美的桌面性能，给您带来高品质体验。

Raspberry Pi 4 B 具备1.5Ghz运行的64位四核处理器，最高支持以60fps 速度刷新的4K分辨率的双显示屏，高达4GB RAM（可根据型号选择1GB、2GB、4GB），2.4/5.0 Ghz 双频无线LAN，蓝牙5.0/BLE，千兆以太网，USB3.0，和PoE功能。

【树莓派两自由度云台】

这是一款专为树莓派设计的两自由度云台扩展板，基于树莓派40PIN GPIO接口设计，直接插入树莓派GPIO口即可使用，同时也将树莓派GPIO口插针引出，不影响树莓派GPIO口的使用。模块采用I2C接口控制，仅需2根信号线即可实现云台转动和光强检测，板载PCA9685芯片，可输出12位分辨率的PWM控制云台转动；板载TSL25911FN，内置ADC，能够得到近似人眼对光的反应，辅助模块工作；板载电平转换电路，兼容3.3V/5V的工作电平；配套专用亚克力板用以固定树莓派摄像头，将摄像头搭配在云台上，拍摄时更方便。
树莓派+mediapipe+pid 摄像头人脸跟随图6

【mediapipe】
MediaPipe 是一款由 Google Research 开发并开源的多媒体机器学习模型应用框架。在谷歌，一系列重要产品，如、Google Lens、ARCore、Google Home 以及，都已深度整合了 MediaPipe。
树莓派+mediapipe+pid 摄像头人脸跟随图7

【安装配置】
1、配置云台
参考：https://wiki.dfrobot.com.cn/FIT0731%20%20%E6%A0%91%E8%8E%93%E6%B4%BE%E4%B8%A4%E8%87%AA%E7%94%B1%E5%BA%A6%E4%BA%91%E5%8F%B0
安装BCM2835库时，将“wget http://www.airspayce.com/mikem/bcm2835/bcm2835-1.68.tar.gz”中修改为：http://www.airspayce.com/mikem/bcm2835/bcm2835-1.71.tar.gz
2、安装mediapipe库
在终端中使用：pip3 install mediapipe-rpi4
树莓派+mediapipe+pid 摄像头人脸跟随图8

【测试云台程序】


#!/usr/bin/python
import time
import RPi.GPIO as GPIO
from PCA9685 import PCA9685

pwm = PCA9685()
try:
    print ("This is an PCA9685 routine")    
    pwm.setPWMFreq(50)
    #pwm.setServoPulse(1,500) 
    #pwm.setRotationAngle(1, 130)
    
    
    # setServoPulse(2,2500)
        
    pwm.setRotationAngle(0, 60)   
            

except:
    pwm.exit_PCA9685()
    print ("\nProgram end")
    exit()

复制代码

【测试mediapipe】


import cv2
import time
import mediapipe as mp
mp_face_detection = mp.solutions.face_detection
mp_drawing = mp.solutions.drawing_utils
pTime = 0
cTime = 0
# For webcam input:
cap = cv2.VideoCapture(0)
with mp_face_detection.FaceDetection(
    model_selection=0, min_detection_confidence=0.8) as face_detection:
  while cap.isOpened():
    success, image = cap.read()
    box=[]
    
    if not success:
      print("Ignoring empty camera frame.")
      # If loading a video, use 'break' instead of 'continue'.
      continue

    # To improve performance, optionally mark the image as not writeable to
    # pass by reference.
    h,w,c=image.shape
    image.flags.writeable = False
    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    results = face_detection.process(image)

    # Draw the face detection annotations on the image.
    image.flags.writeable = True
    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
    if results.detections:
      for detection in results.detections:
        #mp_drawing.draw_detection(image, detection)
        print(detection.location_data.relative_bounding_box)
        box=detection.location_data.relative_bounding_box
        #cx,cy,cw,ch=box
        cx=box.xmin
        cy=box.ymin
        cw=box.width
        ch=box.height
        cv2.rectangle(image, (int(cx*w) , int(cy*h) ), (int((cx+cw)*w) , int((cy+ch)*h)),(0, 255, 0), 2)
    # Flip the image horizontally for a selfie-view display.
    cTime = time.time()
    fps = 1 / (cTime - pTime)
    pTime = cTime

    cv2.putText(image, str(int(fps)), (10, 70), cv2.FONT_HERSHEY_PLAIN, 3,(255, 0, 255), 3)
    cv2.imshow('MediaPipe Face Detection', cv2.flip(image, 1))
    if cv2.waitKey(5) & 0xFF == 27:
      break
cap.release()

复制代码

【完整程序】


#!/usr/bin/python
import time
import RPi.GPIO as GPIO
from PCA9685 import PCA9685
import cv2
import numpy as np
import mediapipe as mp
mp_face_detection = mp.solutions.face_detection
mp_drawing = mp.solutions.drawing_utils
pTime = 0
cTime = 0
# For webcam input:
cap = cv2.VideoCapture(0)
pwm = PCA9685()
pwm.setPWMFreq(50)
pwm.setRotationAngle(1, 110)
pwm.setRotationAngle(0, 90)   
            
kp=0.01
kd=0.01
kp_y=0.02
kd_y=0.01
pre_x=90
pre_y=110  #定义舵机初始角度

pre_err_x=0
pre_err_y=0

with mp_face_detection.FaceDetection(
    model_selection=0, min_detection_confidence=0.65) as face_detection:
  while cap.isOpened():
    success, image = cap.read()
    #image=cv2.resize(image, None, fx=0.5, fy=0.5)
    box=[]
    
    if not success:
      print("Ignoring empty camera frame.")
      # If loading a video, use 'break' instead of 'continue'.
      continue

    # To improve performance, optionally mark the image as not writeable to
    # pass by reference.
    h,w,c=image.shape
    image.flags.writeable = False
    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    results = face_detection.process(image)

    # Draw the face detection annotations on the image.
    image.flags.writeable = True
    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
    if results.detections:
      for detection in results.detections:
        #mp_drawing.draw_detection(image, detection)
        print(detection.location_data.relative_bounding_box)
        box=detection.location_data.relative_bounding_box
        #cx,cy,cw,ch=box
        cx=box.xmin
        cy=box.ymin
        cw=box.width
        ch=box.height
        cv2.circle(image, (int(cx*w), int(cy*h)), 5, (255, 0, 255), cv2.FILLED)
        cv2.rectangle(image, (int(cx*w) , int(cy*h) ), (int((cx+cw)*w) , int((cy+ch)*h)),(0, 255, 0), 2)
        
        err_x=(w/2-cx*w)
        err_y=(cy*h+ch*h-h/2)
        dis_err_x=err_x-pre_err_x
        dis_err_y=err_y-pre_err_y
        
        
  
        pre_x=(pre_x+kp*err_x+kd*dis_err_x)
        pre_y=(pre_y+kp_y*err_y+kd_y*dis_err_y)
        pre_err_x=err_x
        pre_err_y=err_y
        print(h,"  ",pre_y," ",int(cy*h),"  ",err_y,"  ")
            #防止超调
        if pre_x<0:
               pre_x=0
        elif pre_x>180:
               pre_x=180
               
        if pre_y<90:
               pre_y=90
        elif pre_y>130:
               pre_y=130
        pwm.setRotationAngle(0, pre_x)
        pwm.setRotationAngle(1, pre_y)
           
        
    # Flip the image horizontally for a selfie-view display.
    cTime = time.time()
    fps = 1 / (cTime - pTime)
    pTime = cTime

    cv2.putText(image, str(int(fps)), (10, 70), cv2.FONT_HERSHEY_PLAIN, 3,(255, 0, 255), 3)
    cv2.imshow('MediaPipe Face Detection', cv2.flip(image, 1))
    if cv2.waitKey(5) & 0xFF == 27:
      pwm.exit_PCA9685()
      break
cap.release()

复制代码