jetson nano 人脸识别（Paul McWhorter课程）

1、配置环境

opencv 4.1.1、python3.6.9、nuimpy1.16.1、

扩大内存 ，tx2中修改mem是一样的

1、下载

git clone https://github.com/JetsonHacksNano/installSwapfile

2、安装

 ./installSwapfile/installSwapfile.sh

3、重启

2、下载人脸识别包dlib 和 face-recognition

无法安装人脸检测dlib库的解决方法_dlib安装失败-优快云博客

Index of /files

验证是否安装成功

python3
import dlib
dlib.DLIB_USE_CUDA

3、训练与测试

训练数据链接：https://pan.baidu.com/s/1Dr--LZN3-SD7e8LdfJTBpw?pwd=q2t1 
提取码：q2t1

训练+单张测试

import face_recognition
import cv2
import os

Encodings=[]
Names=[]


image_dir = '/home/nano/faceRecognizer/known'
#遍历文件夹里面文件，进行训练并存储对应的编码数据
for root, dirs, files in os.walk(image_dir):
    print(files)#得到文件夹里所有图片并生成一个数组
    for file in files:#遍历数组
        path = os.path.join(root, file)#得到训练图片的根目录
        print(path)
        name = os.path.splitext(file)[0]
        print(name)
        person = face_recognition.load_image_file(path)
        encoding = face_recognition.face_encodings(person)[0]
        Encodings.append(encoding)
        Names.append(name)
print(Names)

#定义测试图片及输出文字字体
font = cv2.FONT_HERSHEY_SIMPLEX
testImage=face_recognition.load_image_file('/home/nano/faceRecognizer/unknown/u12.jpg')
#找到目标位置
facePositions=face_recognition.face_locations(testImage)
#将测试的所有图片进行编码
allEncodings=face_recognition.face_encodings(testImage,facePositions)
#将测试的的rgb图像转成bgr
testImage=cv2.cvtColor(testImage,cv2.COLOR_RGB2BGR)
#匹配数据库是不是有测试图片相同的编码
for (top,right,bottom,left),face_encoding in zip(facePositions,allEncodings):
    name='Unknown Person'
    matches=face_recognition.compare_faces(Encodings,face_encoding)
    if True in matches:
        first_match_index=matches.index(True)#找到匹配的索引
        name=Names[first_match_index]#用找到的索引传给name列表，找到人名
    cv2.rectangle(testImage,(left,top),(right,bottom),(0,0,255),2)
    cv2.putText(testImage,name,(left,top-6),font,.75,(0,255,255),2)

cv2.imshow('mywindow',testImage)

if cv2.waitKey(0)==ord('q'):
    cv2.destroyAllWindows()

训练+检测文件夹所有目标 

import face_recognition
import cv2
import os

Encodings=[]
Names=[]


image_dir = '/home/nano/faceRecognizer/known'
#遍历文件夹里面文件，进行训练并存储对应的编码数据
for root, dirs, files in os.walk(image_dir):
    print(files)#得到文件夹里所有图片并生成一个数组
    for file in files:#遍历数组
        path = os.path.join(root, file)#得到训练图片的根目录
        print(path)
        name = os.path.splitext(file)[0]
        print(name)
        person = face_recognition.load_image_file(path)
        encoding = face_recognition.face_encodings(person)[0]
        Encodings.append(encoding)
        Names.append(name)
print(Names)

#定义测试图片及输出文字字体
font = cv2.FONT_HERSHEY_SIMPLEX

for root,dirs,files in os.walk(image_dir):
    for file in files:
        print(root)
        print(file)
        testImagePath=os.path.join(root,file)
        testImage=face_recognition.load_image_file(testImagePath)
        #testImage=face_recognition.load_image_file('/home/nano/faceRecognizer/unknown/u12.jpg')
        #找到目标位置
        facePositions=face_recognition.face_locations(testImage)
        #将测试的所有图片进行编码
        allEncodings=face_recognition.face_encodings(testImage,facePositions)
        #将测试的的rgb图像转成bgr
        testImage=cv2.cvtColor(testImage,cv2.COLOR_RGB2BGR)
        #匹配数据库是不是有测试图片相同的编码
        for (top,right,bottom,left),face_encoding in zip(facePositions,allEncodings):
            name='Unknown Person'
            matches=face_recognition.compare_faces(Encodings,face_encoding)
            if True in matches:
                first_match_index=matches.index(True)#找到匹配的索引
                name=Names[first_match_index]#用找到的索引传给name列表，找到人名
            cv2.rectangle(testImage,(left,top),(right,bottom),(0,0,255),2)
            cv2.putText(testImage,name,(left,top-6),font,.75,(0,255,255),2)

        cv2.imshow('mywindow',testImage)

        if cv2.waitKey(0)==ord('q'):
            cv2.destroyAllWindows()

4、用pickle存储训练好的人脸模型

4.1、train-save.py 训练+保存

import face_recognition
import cv2
import os
import pickle

Encodings=[]
Names=[]


image_dir = '/home/nano/faceRecognizer/known'
#遍历文件夹里面文件，进行训练并存储对应的编码数据
for root, dirs, files in os.walk(image_dir):
    print(files)#得到文件夹里所有图片并生成一个数组
    for file in files:#遍历数组
        path = os.path.join(root, file)#得到训练图片的根目录
        print(path)
        name = os.path.splitext(file)[0]
        print(name)
        person = face_recognition.load_image_file(path)
        encoding = face_recognition.face_encodings(person)[0]
        Encodings.append(encoding)
        Names.append(name)
print(Names)

#使用pickle存储训练好的识别模型
with open('train.pkl','wb') as f:
    pickle.dump(Names,f)
    pickle.dump(Encodings,f)

4.2、read-recognize.py调用保存的pickle模型进行人脸识别

import face_recognition
import cv2
import os
import pickle



Encodings=[]
Names=[]

with open('train.pkl','rb') as f:
    Names=pickle.load(f)
    Encodings=pickle.load(f)

#定义测试图片及输出文字字体
font = cv2.FONT_HERSHEY_SIMPLEX

image_dir = '/home/nano/faceRecognizer/unknown'
for root,dirs,files in os.walk(image_dir):
    for file in files:
        print(root)
        print(file)
        testImagePath=os.path.join(root,file)
        testImage=face_recognition.load_image_file(testImagePath)
        #testImage=face_recognition.load_image_file('/home/nano/faceRecognizer/unknown/u12.jpg')
        #找到目标位置
        facePositions=face_recognition.face_locations(testImage)
        #将测试的所有图片进行编码
        allEncodings=face_recognition.face_encodings(testImage,facePositions)
        #将测试的的rgb图像转成bgr
        testImage=cv2.cvtColor(testImage,cv2.COLOR_RGB2BGR)
        #匹配数据库是不是有测试图片相同的编码
        for (top,right,bottom,left),face_encoding in zip(facePositions,allEncodings):
            name='Unknown Person'
            matches=face_recognition.compare_faces(Encodings,face_encoding)
            if True in matches:
                first_match_index=matches.index(True)#找到匹配的索引
                name=Names[first_match_index]#用找到的索引传给name列表，找到人名
            cv2.rectangle(testImage,(left,top),(right,bottom),(0,0,255),2)
            cv2.putText(testImage,name,(left,top-6),font,.75,(0,255,255),2)

        cv2.imshow('mywindow',testImage)
        if cv2.waitKey(0)==ord('q'):
            cv2.destroyAllWindows()

结果

5、调用摄像头人脸识别

帧率8~9，不过识别挺准的

import face_recognition
import cv2
import os
import pickle
import time


fpsReport=0
scaleFactor=.2#图像缩小比例，越小检测速度越快


Encodings=[]
Names=[]

with open('train.pkl','rb') as f:
    Names=pickle.load(f)
    Encodings=pickle.load(f)

#定义测试图片及输出文字字体
font = cv2.FONT_HERSHEY_SIMPLEX

cam = cv2.VideoCapture(0)
timeStamp=time.time()

while True:
    _,frame = cam.read()
    #图片缩小scalefactor，加快检测速度
    frameSmall = cv2.resize(frame, (0, 0), fx=scaleFactor, fy=scaleFactor)
    frameRGB = cv2.cvtColor(frameSmall, cv2.COLOR_BGR2RGB)
    facePositions = face_recognition.face_locations(frameRGB, model='cnn')
    allEncodings = face_recognition.face_encodings(frameRGB, facePositions)
    for (top, right, bottom, left), face_encoding in zip(facePositions, allEncodings):
        name = 'Unkown Person'
        matches = face_recognition.compare_faces(Encodings, face_encoding)
        if True in matches:
            first_match_index = matches.index(True)
            name = Names[first_match_index]
            #将检测框放大scalefactor倍，上面图片是缩小三倍检测的结果，检测框也是缩小三倍的
        top = int(top / scaleFactor)
        right = int(right / scaleFactor)
        bottom = int(bottom / scaleFactor)
        left = int(left / scaleFactor)

        cv2.rectangle(frame,(left,top),(right,bottom),(0, 0, 255), 2)
        cv2.putText(frame, name, (left, top - 6), font, .75, (0, 0, 255), 2)

    #计算帧率并显示在右上角
    dt = time.time() - timeStamp
    fps = 1 / dt
    fps = round(fps,2)
    #低通滤波，避免fps反复刷新消除噪音
    fpsReport = .90 * fpsReport + .1 * fps
    # print('fps is:',round(fpsReport,1))
    timeStamp = time.time()
    cv2.rectangle(frame, (0, 0), (100, 40), (0, 0, 255), -1)
    cv2.putText(frame, str(round(fpsReport, 1)) + 'fps', (0, 25), font, .75, (0, 255, 255, 2))

    cv2.imshow("cam",frame)
    if cv2.waitKey(1)==ord('q'):
        break
cam.release()
cv2.destroyAllWindows()