NVIDIA | Technology Tutorials

<span data-mce-type="bookmark" style="display: inline-block; width: 0px; overflow: hidden; line-height: 0;" class="mce_SELRES_start"></span> In this video lesson we should you a simple method to train our face recognizer on larger data sets. We use the python os.walk command to step through, and train automatically on all the training images in our folder. We then show how to store our training set to our SD card using the pickle utility. This allows us to train once, and use the trained model over and over.

For your convenience, the code below is what we developed to allow training our face recognition model.

import face_recognition
import cv2
import os
import pickle
print(cv2.__version__)

Encodings=[]
Names=[]
j=0

image_dir='/home/pjm/Desktop/pyPro/demoimages/known'
for root, dirs, files in os.walk(image_dir):
    print(files)
    for file in files:
        fullPath=os.path.join(root,file)
        print(fullPath)
        name=os.path.splitext(file)[0]
        print(name)
        person=face_recognition.load_image_file(fullPath)
        encoding=face_recognition.face_encodings(person)[0]
        Encodings.append(encoding)
        Names.append(name)
print(Names)
with open('train.pkl','wb') as f:
        pickle.dump(Names,f)
        pickle.dump(Encodings,f)

import face_recognition

import cv2

import os

import pickle

print(cv2.__version__)

Encodings=[]

Names=[]

j=0

image_dir='/home/pjm/Desktop/pyPro/demoimages/known'

for root, dirs, files in os.walk(image_dir):

print(files)

for file in files:

fullPath=os.path.join(root,file)

print(fullPath)

name=os.path.splitext(file)[0]

print(name)

person=face_recognition.load_image_file(fullPath)

encoding=face_recognition.face_encodings(person)[0]

Encodings.append(encoding)

Names.append(name)

print(Names)

with open('train.pkl','wb') as f:

pickle.dump(Names,f)

pickle.dump(Encodings,f)

Then this is a simple program that loads the trained model, and uses it to recognize people in unknown images.

import cv2
print(cv2.__version__)
import face_recognition
import pickle

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

font=cv2.FONT_HERSHEY_SIMPLEX
testImage=face_recognition.load_image_file('/home/pjm/Desktop/pyPro/demoimages/unkno$

facePositions=face_recognition.face_locations(testImage)
allEncodings=face_recognition.face_encodings(testImage,facePositions)

testImage=cv2.cvtColor(testImage,cv2.COLOR_RGB2BGR)
for (top,right,bottom,left), face_encoding in zip(facePositions, allEncodings):
    name='Unknown Life Form'
    matches=face_recognition.compare_faces(Encodings,face_encoding)
    if True in matches:
        first_match_index=matches.index(True)
        name=Names[first_match_index]
    cv2.rectangle(testImage,(left,top),(right,bottom),(255,0,0),2)
    cv2.rectangle(testImage, (left,top),(left+200, top+30),(0,255,255),-1)
    cv2.putText(testImage,name,(left,top+20),font,.75,(255,0,0),2)
cv2.imshow('mywindow',testImage)
cv2.moveWindow('mywindow',0,0)
if cv2.waitKey(0)==ord('q'):
    cv2.destroyAllWindows()

import cv2

print(cv2.__version__)

import face_recognition

import pickle

with open('train.pkl','rb') as f:

Names=pickle.load(f)

Encodings=pickle.load(f)

font=cv2.FONT_HERSHEY_SIMPLEX

testImage=face_recognition.load_image_file('/home/pjm/Desktop/pyPro/demoimages/unkno$

facePositions=face_recognition.face_locations(testImage)

allEncodings=face_recognition.face_encodings(testImage,facePositions)

testImage=cv2.cvtColor(testImage,cv2.COLOR_RGB2BGR)

for (top,right,bottom,left), face_encoding in zip(facePositions, allEncodings):

name='Unknown Life Form'

matches=face_recognition.compare_faces(Encodings,face_encoding)

if True in matches:

first_match_index=matches.index(True)

name=Names[first_match_index]

cv2.rectangle(testImage,(left,top),(right,bottom),(255,0,0),2)

cv2.rectangle(testImage, (left,top),(left+200, top+30),(0,255,255),-1)

cv2.putText(testImage,name,(left,top+20),font,.75,(255,0,0),2)

cv2.imshow('mywindow',testImage)

cv2.moveWindow('mywindow',0,0)

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

cv2.destroyAllWindows()

In this Video Tutorial we show how a camera on a pan/tilt control system can be programmed to search for an object of interest, and then track it when found. Our system has two independent camera systems, and each can track a separate item of interest independently. The code is written in python, using the OpenCV library. The video takes you through the lesson step-by-step, and then the code is included below for your convenience.

If you want to play along at home, we are using the Jetson Xavier NX, which you can pick up HERE. You will also need to of the bracket/servo kits, which you can get HERE, and then two Raspberry Pi Version two cameras, available HERE.

import cv2
import numpy as np
import time
from adafruit_servokit import ServoKit
print(cv2.__version__)
timeMark=time.time()
dtFIL=0

scanRight=True
scanLeft=True

def nothing(x):
    pass

cv2.namedWindow('TrackBars')
cv2.moveWindow('TrackBars',1320,0)
cv2.createTrackbar('hueLower', 'TrackBars',100,179,nothing)
cv2.createTrackbar('hueUpper', 'TrackBars',116,179,nothing)
cv2.createTrackbar('satLow', 'TrackBars',160,255,nothing)
cv2.createTrackbar('satHigh', 'TrackBars',255,255,nothing)
cv2.createTrackbar('valLow', 'TrackBars',150,255,nothing)
cv2.createTrackbar('valHigh', 'TrackBars',255,255,nothing)

cv2.namedWindow('TrackBars2')
cv2.moveWindow('TrackBars2',1100,0)
cv2.createTrackbar('hueLower2', 'TrackBars2',150,179,nothing)
cv2.createTrackbar('hueUpper2', 'TrackBars2',170,179,nothing)
cv2.createTrackbar('satLow2', 'TrackBars2',160,255,nothing)
cv2.createTrackbar('satHigh2', 'TrackBars2',255,255,nothing)
cv2.createTrackbar('valLow2', 'TrackBars2',150,255,nothing)
cv2.createTrackbar('valHigh2', 'TrackBars2',255,255,nothing)


kit=ServoKit(channels=16)

tilt1=90
pan1=90
tilt2=90
pan2=90

dPan1=1
dPan2=1
dTilt1=10
dTilt2=10

kit.servo[0].angle=pan1
kit.servo[1].angle=tilt1
kit.servo[2].angle=pan2
kit.servo[3].angle=tilt2

width=720
height=480
flip=2
font=cv2.FONT_HERSHEY_SIMPLEX
camSet1='nvarguscamerasrc sensor-id=0 ee-mode=1 ee-strength=0 tnr-mode=2 tnr-strength=1 wbmode=3 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! videobalance contrast=1.3 brightness=-.2 saturation=1.2 ! appsink drop=True'
camSet2='nvarguscamerasrc sensor-id=1 ee-mode=1 ee-strength=0 tnr-mode=2 tnr-strength=1 wbmode=3 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! videobalance contrast=1.3 brightness=-.2 saturation=1.2 ! appsink drop=True'

#camSet='nvarguscamerasrc sensor-id=0 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! appsink'
#camSet ='v4l2src device=/dev/video1 ! video/x-raw,width='+str(width)+',height='+str(height)+',framerate=20/1 ! videoconvert ! appsink'
cam1=cv2.VideoCapture(camSet1)
cam2=cv2.VideoCapture(camSet2)
while True:
    _, frame1 = cam1.read()
    _, frame2 = cam2.read()

    hsv1=cv2.cvtColor(frame1,cv2.COLOR_BGR2HSV)
    hsv2=cv2.cvtColor(frame2,cv2.COLOR_BGR2HSV)

    hueLow=cv2.getTrackbarPos('hueLower', 'TrackBars')
    hueUp=cv2.getTrackbarPos('hueUpper', 'TrackBars')

    Ls=cv2.getTrackbarPos('satLow', 'TrackBars')
    Us=cv2.getTrackbarPos('satHigh', 'TrackBars')

    Lv=cv2.getTrackbarPos('valLow', 'TrackBars')
    Uv=cv2.getTrackbarPos('valHigh', 'TrackBars')

    l_b=np.array([hueLow,Ls,Lv])
    u_b=np.array([hueUp,Us,Uv])


    hueLow2=cv2.getTrackbarPos('hueLower2', 'TrackBars2')
    hueUp2=cv2.getTrackbarPos('hueUpper2', 'TrackBars2')

    Ls2=cv2.getTrackbarPos('satLow2', 'TrackBars2')
    Us2=cv2.getTrackbarPos('satHigh2', 'TrackBars2')

    Lv2=cv2.getTrackbarPos('valLow2', 'TrackBars2')
    Uv2=cv2.getTrackbarPos('valHigh2', 'TrackBars2')

    l_b2=np.array([hueLow2,Ls2,Lv2])
    u_b2=np.array([hueUp2,Us2,Uv2])

    FGmask1=cv2.inRange(hsv1,l_b,u_b)
    FGmask2=cv2.inRange(hsv2,l_b2,u_b2)

    cv2.imshow('FGmask1',FGmask1)
    cv2.moveWindow('FGmask1',0,0)

    cv2.imshow('FGmask2',FGmask2)
    cv2.moveWindow('FGmask2',350,0)

    contours1,_ = cv2.findContours(FGmask1,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
    contours1=sorted(contours1,key=lambda x:cv2.contourArea(x),reverse=True)
    for cnt in contours1:
        area=cv2.contourArea(cnt)
        (x,y,w,h)=cv2.boundingRect(cnt)
        if area>=100:
            scanLeft=False
            cv2.rectangle(frame1,(x,y),(x+w,y+h),(0,255,255),3)
            objX=x+w/2
            objY=y+h/2
            errorPan1=objX-width/2
            errorTilt1=objY-height/2
            if abs(errorPan1)>15:
                pan1=pan1+errorPan1/40
            if abs(errorTilt1)>15:
                tilt1=tilt1-errorTilt1/40
            if pan1>180:
                pan1=180
                print('Pan Out of Range')
            if pan1<0:
                pan1=0
                print('Pan Out of Range')
            if tilt1>180:
                tilt1=180
                print('Tilt Out of Range')
            if tilt1<0:
                tilt1=0
            kit.servo[2].angle=pan1
            kit.servo[3].angle=tilt1
            break
    
    contours2,_ = cv2.findContours(FGmask2,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
    contours2=sorted(contours2,key=lambda x:cv2.contourArea(x),reverse=True)
    for cnt in contours2:
        area=cv2.contourArea(cnt)
        (x,y,w,h)=cv2.boundingRect(cnt)
        if area>=100:
            scanRight=False
            cv2.rectangle(frame2,(x,y),(x+w,y+h),(0,255,255),3)
            objX=x+w/2
            objY=y+h/2
            errorPan2=objX-width/2
            errorTilt2=objY-height/2
            if abs(errorPan2)>15:
                pan2=pan2+errorPan2/40
            if abs(errorTilt2)>15:
                tilt2=tilt2-errorTilt2/40
            if pan2>180:
                pan2=180
                print('Pan Out of Range')
            if pan2<0:
                pan2=0
                print('Pan Out of Range')
            if tilt2>180:
                tilt2=180
                print('Tilt Out of Range')
            if tilt2<0:
                tilt2=0
            kit.servo[0].angle=pan2
            kit.servo[1].angle=tilt2            
            break
    if scanLeft==True:
        if pan1>=179:
            dPan1=abs(dPan1)*(-1)
        if pan1<=1:
            dPan1=abs(dPan1)
        if pan1>=179 or pan1<=1:
            if tilt1>=170:
                dTilt1=abs(dTilt1)*(-1)
            if tilt1<=10:
                dTilt1=abs(dTilt1)
            tilt1=tilt1+dTilt1
        pan1=pan1+dPan1
        kit.servo[2].angle=pan1
        kit.servo[3].angle=tilt1
    scanLeft=True

    if scanRight==True:
        if pan2>=179:
            dPan2=abs(dPan2)*(-1)
        if pan2<=1:
            dPan2=abs(dPan2)
        if pan2>=179 or pan2<=1:
            if tilt2>=170:
                dTilt2=abs(dTilt2)*(-1)
            if tilt2<=10:
                dTilt2=abs(dTilt2)
            tilt2=tilt2+dTilt2
        pan2=pan2+dPan2
        kit.servo[0].angle=pan2
        kit.servo[1].angle=tilt2
    scanRight=True
    
    frame3=np.hstack((frame1,frame2))
    dt=time.time()-timeMark
    timeMark=time.time()
    dtFIL=.9*dtFIL + .1*dt
    fps=1/dtFIL
    cv2.rectangle(frame3,(0,0),(150,40),(0,0,255),-1)
    cv2.putText(frame3,'fps: '+str(round(fps,1)),(0,30),font,1,(0,255,255),2)

    #cv2.imshow('myCam1',frame1)
    #cv2.imshow('myCam2',frame2)
    cv2.imshow('comboCam',frame3)
    cv2.moveWindow('comboCam',0,450)

    if cv2.waitKey(1)==ord('q'):
        break
cam1.release()
cam2.release()
cv2.destroyAllWindows()

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

import cv2

import numpy as np

import time

from adafruit_servokit import ServoKit

print(cv2.__version__)

timeMark=time.time()

dtFIL=0

scanRight=True

scanLeft=True

def nothing(x):

pass

cv2.namedWindow('TrackBars')

cv2.moveWindow('TrackBars',1320,0)

cv2.createTrackbar('hueLower', 'TrackBars',100,179,nothing)

cv2.createTrackbar('hueUpper', 'TrackBars',116,179,nothing)

cv2.createTrackbar('satLow', 'TrackBars',160,255,nothing)

cv2.createTrackbar('satHigh', 'TrackBars',255,255,nothing)

cv2.createTrackbar('valLow', 'TrackBars',150,255,nothing)

cv2.createTrackbar('valHigh', 'TrackBars',255,255,nothing)

cv2.namedWindow('TrackBars2')

cv2.moveWindow('TrackBars2',1100,0)

cv2.createTrackbar('hueLower2', 'TrackBars2',150,179,nothing)

cv2.createTrackbar('hueUpper2', 'TrackBars2',170,179,nothing)

cv2.createTrackbar('satLow2', 'TrackBars2',160,255,nothing)

cv2.createTrackbar('satHigh2', 'TrackBars2',255,255,nothing)

cv2.createTrackbar('valLow2', 'TrackBars2',150,255,nothing)

cv2.createTrackbar('valHigh2', 'TrackBars2',255,255,nothing)

kit=ServoKit(channels=16)

tilt1=90

pan1=90

tilt2=90

pan2=90

dPan1=1

dPan2=1

dTilt1=10

dTilt2=10

kit.servo[0].angle=pan1

kit.servo[1].angle=tilt1

kit.servo[2].angle=pan2

kit.servo[3].angle=tilt2

width=720

height=480

flip=2

font=cv2.FONT_HERSHEY_SIMPLEX

camSet1='nvarguscamerasrc sensor-id=0 ee-mode=1 ee-strength=0 tnr-mode=2 tnr-strength=1 wbmode=3 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! videobalance contrast=1.3 brightness=-.2 saturation=1.2 ! appsink drop=True'

camSet2='nvarguscamerasrc sensor-id=1 ee-mode=1 ee-strength=0 tnr-mode=2 tnr-strength=1 wbmode=3 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! videobalance contrast=1.3 brightness=-.2 saturation=1.2 ! appsink drop=True'

#camSet='nvarguscamerasrc sensor-id=0 ! video/x-raw(memory:NVMM), width=3264, height=2464, framerate=21/1,format=NV12 ! nvvidconv flip-method='+str(flip)+' ! video/x-raw, width='+str(width)+', height='+str(height)+', format=BGRx ! videoconvert ! video/x-raw, format=BGR ! appsink'

#camSet ='v4l2src device=/dev/video1 ! video/x-raw,width='+str(width)+',height='+str(height)+',framerate=20/1 ! videoconvert ! appsink'

cam1=cv2.VideoCapture(camSet1)

cam2=cv2.VideoCapture(camSet2)

while True:

_, frame1 = cam1.read()

_, frame2 = cam2.read()

hsv1=cv2.cvtColor(frame1,cv2.COLOR_BGR2HSV)

hsv2=cv2.cvtColor(frame2,cv2.COLOR_BGR2HSV)

hueLow=cv2.getTrackbarPos('hueLower', 'TrackBars')

hueUp=cv2.getTrackbarPos('hueUpper', 'TrackBars')

Ls=cv2.getTrackbarPos('satLow', 'TrackBars')

Us=cv2.getTrackbarPos('satHigh', 'TrackBars')

Lv=cv2.getTrackbarPos('valLow', 'TrackBars')

Uv=cv2.getTrackbarPos('valHigh', 'TrackBars')

l_b=np.array([hueLow,Ls,Lv])

u_b=np.array([hueUp,Us,Uv])

hueLow2=cv2.getTrackbarPos('hueLower2', 'TrackBars2')

hueUp2=cv2.getTrackbarPos('hueUpper2', 'TrackBars2')

Ls2=cv2.getTrackbarPos('satLow2', 'TrackBars2')

Us2=cv2.getTrackbarPos('satHigh2', 'TrackBars2')

Lv2=cv2.getTrackbarPos('valLow2', 'TrackBars2')

Uv2=cv2.getTrackbarPos('valHigh2', 'TrackBars2')

l_b2=np.array([hueLow2,Ls2,Lv2])

u_b2=np.array([hueUp2,Us2,Uv2])

FGmask1=cv2.inRange(hsv1,l_b,u_b)

FGmask2=cv2.inRange(hsv2,l_b2,u_b2)

cv2.imshow('FGmask1',FGmask1)

cv2.moveWindow('FGmask1',0,0)

cv2.imshow('FGmask2',FGmask2)

cv2.moveWindow('FGmask2',350,0)

contours1,_ = cv2.findContours(FGmask1,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)

contours1=sorted(contours1,key=lambda x:cv2.contourArea(x),reverse=True)

for cnt in contours1:

area=cv2.contourArea(cnt)

(x,y,w,h)=cv2.boundingRect(cnt)

if area>=100:

scanLeft=False

cv2.rectangle(frame1,(x,y),(x+w,y+h),(0,255,255),3)

objX=x+w/2

objY=y+h/2

errorPan1=objX-width/2

errorTilt1=objY-height/2

if abs(errorPan1)>15:

pan1=pan1+errorPan1/40

if abs(errorTilt1)>15:

tilt1=tilt1-errorTilt1/40

if pan1>180:

pan1=180

print('Pan Out of Range')

if pan1<0:

pan1=0

print('Pan Out of Range')

if tilt1>180:

tilt1=180

print('Tilt Out of Range')

if tilt1<0:

tilt1=0

kit.servo[2].angle=pan1

kit.servo[3].angle=tilt1

break

contours2,_ = cv2.findContours(FGmask2,cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)

contours2=sorted(contours2,key=lambda x:cv2.contourArea(x),reverse=True)

for cnt in contours2:

area=cv2.contourArea(cnt)

(x,y,w,h)=cv2.boundingRect(cnt)

if area>=100:

scanRight=False

cv2.rectangle(frame2,(x,y),(x+w,y+h),(0,255,255),3)

objX=x+w/2

objY=y+h/2

errorPan2=objX-width/2

errorTilt2=objY-height/2

if abs(errorPan2)>15:

pan2=pan2+errorPan2/40

if abs(errorTilt2)>15:

tilt2=tilt2-errorTilt2/40

if pan2>180:

pan2=180

print('Pan Out of Range')

if pan2<0:

pan2=0

print('Pan Out of Range')

if tilt2>180:

tilt2=180

print('Tilt Out of Range')

if tilt2<0:

tilt2=0

kit.servo[0].angle=pan2

kit.servo[1].angle=tilt2

break

if scanLeft==True:

if pan1>=179:

dPan1=abs(dPan1)*(-1)

if pan1<=1:

dPan1=abs(dPan1)

if pan1>=179 or pan1<=1:

if tilt1>=170:

dTilt1=abs(dTilt1)*(-1)

if tilt1<=10:

dTilt1=abs(dTilt1)

tilt1=tilt1+dTilt1

pan1=pan1+dPan1

kit.servo[2].angle=pan1

kit.servo[3].angle=tilt1

scanLeft=True

if scanRight==True:

if pan2>=179:

dPan2=abs(dPan2)*(-1)

if pan2<=1:

dPan2=abs(dPan2)

if pan2>=179 or pan2<=1:

if tilt2>=170:

dTilt2=abs(dTilt2)*(-1)

if tilt2<=10:

dTilt2=abs(dTilt2)

tilt2=tilt2+dTilt2

pan2=pan2+dPan2

kit.servo[0].angle=pan2

kit.servo[1].angle=tilt2

scanRight=True

frame3=np.hstack((frame1,frame2))

dt=time.time()-timeMark

timeMark=time.time()

dtFIL=.9*dtFIL + .1*dt

fps=1/dtFIL

cv2.rectangle(frame3,(0,0),(150,40),(0,0,255),-1)

cv2.putText(frame3,'fps: '+str(round(fps,1)),(0,30),font,1,(0,255,255),2)

#cv2.imshow('myCam1',frame1)

#cv2.imshow('myCam2',frame2)

cv2.imshow('comboCam',frame3)

cv2.moveWindow('comboCam',0,450)

if cv2.waitKey(1)==ord('q'):

break

cam1.release()

cam2.release()

cv2.destroyAllWindows()

Technology Tutorials

Tag Archives: NVIDIA

AI on the Jetson Nano LESSON 57: Read a Push Button Switch on the GPIO Pins With Pull Up Resistors

Jetson Xavier NX Lesson 15: Training the Face Recognition Program to Recognize People

AI on the Jetson Nano LESSON 56: Using the GPIO Pins on the Jetson Nano

Jetson Xavier NX Lesson 13: Installing Face Recognition and Identification Libraries

Jetson Xavier NX Lesson 12: Intelligent Scanning for Objects of Interest

Making The World a Better Place One High Tech Project at a Time. Enjoy!