ESP-Cam: How do I find out the MATCH ID locally on the camera if a face recognition was successful?

[PeterN]

Hi CiccioCB 

I am experimenting with ESP32_CAM and M5-CAMERA, and I really like it !!!.

  

I use an M5Stack (and your ESP32_dev_board) with its TFT-display as a client running the example code. On the client screen I can mark a recognized face with a rectangle and can print its MATCH_ID which is filtered with TFT.CAMERA$  from the HTTP-header..

But  how can I use the last MATCH_ID on the CAMERA as well?  I do  miss there a variable or an event to get the last face recognition MATCH_ID (and as well the coordinates of the face detection).

Did  I oversee something?  

PeterN 

[cicciocb]

Hi Peter,

sorry for the late answer.

Just found that I forgot to document the function CAMERA.DETECTED$.

This function returns the same information sent in the header.

I'll update the help at the same time for the next release

[PeterN]

Good morning ciccioCB

Fantastic! That is what I hoped for! 

Thanks!!

I played a bit with my M5 devices and ANNEX-Cam:

https://vimeo.com/466643953

PS: I could not post a picture or append a file here, because the server refuses the upload.

[cicciocb]

Hi Peter,

I forgot to add that there is also another associated event, ONCAMDETECT

[PeterN]

That really completes the usability!

[PeterN]

Good evening CiccioCB,

I did some successfull testing with face recognition  on the M5camera 

- without a separate client (= other ANNEX32 or a browser-script) looking at the cam-picture and its header 

- but locally using the  OnCamDetect event and then the CAMERA.DETECTED$.

It seems to be a bit tricky to avoid  to have to many detection events, if the cam sees a face. This caused a lot of unwanted freezes and reboots until I temporarily deactivated the face detection to give time to  process the needed actions.   

If someone likes to try that maybe my code can help as a start.

This was the code for my experiment with the M5-camera looking at a set of fotos in a diashow-window at my notebook:

CODE: xxxx.bas

---

' ##################################################################################
' M5Camera (or ESP32-CAM)  for face recognition 
' Saves the last 7  detected faces in /faces.fh
' Ccompare the currently detected face with the stored fraces
'
'###################################################################################
rate = 10
detection_msg$ = "----"

'The Camera can recognize max 7 faceIDs and stored them in /faces.fh
dim FACE$(7)="PeterN","MonikaN","Obama","Johnson","Trump","Merkel","Face #7"
'dim FACE$(7)="Face #1","Face #2","Face #3","Face #4","Face #5","Face #6","Face #7"

' must set the resolution to 320*240 for facerecognition
print camera.setup(4) '320*240
print camera.params("vflip", 1)
print camera.params("hmirror", 1)

print camera.params("detection",1)
print camera.params("readfaces",1)
print camera.params("recognition",1)

'################################################
OnCamDetect FACE_DETECTED
onHtmlReload setpage
gosub setpage

wait

'###################################################################################
'###################################################################################

setpage:
'----------
cls
a$ = ""
A$ = A$ + |<h2> AnnexCAM Face Recognition</h2>|
A$ = A$ + |<img id='camera' src="picture" alt="AnnexCam" style="width:auto;"><br><br>|
A$ = A$ + "<br><br>"
A$ = A$ + button$("V Flip", v_flip) + button$("H mirror", h_mirror)
A$ = A$ + "<br><br>"
A$ = A$ + button$("ENROLL THE NEXT FACE", ENROLL_FACES) + button$("Save all Faces to /faces.fh", SAVE_FACES)
A$ = A$ + "<br><br>"
A$ = A$ + textbox$(detection_msg$) 

html a$
jscall "set_pictimer(" + str$(rate) + ");"
'autorefresh 250
autorefresh 1000
a$ = ""
return

' ##################################################################################
ENROLL_FACES:
'----------
print camera.params("enroll", 1)
return

'###################################################################################
SAVE_FACES:
'----------
print camera.params("savefaces", 1)
return

'###################################################################################
v_flip:
'----------
v = camera.getvalue("vflip")
print camera.params("vflip", 1 - v)
return

'###################################################################################
h_mirror:
'----------
v = camera.getvalue("hmirror")
print camera.params("hmirror", 1 - v)
return

'###################################################################################

FACE_DETECTED:
'----------
detection_msg$ = Camera.detected$
' Temporarily deactivate the face detection to prevent getting to many detection-events
print camera.params("detection",0)  

'The detected face may be one of the 7 faces stored in /face.fh 
detection_msg$ = word$(detection_msg$, 2, "Match ID:") 
if detection_msg$ <> "" then detection_msg$ = FACE$(val(detection_msg$)) else detection_msg$ = "?????"

refresh
pause 1000

'reactivate face detection  
print camera.params("detection",1)

return

[PeterN]

Hi CiccioCB,

I am still very enthusiastic about ANNEXCAM. 

Could you give more detailed information about the functions that had to be removed for the sake of speed and memory?   

Thanks 

Peter

[cicciocb]

Hi Peter,

this is the list of the functionalities included / removed in the ANNEXCAM.

This table will be included in the documentation.

Functionalities enabled in the ANNEXCAM version

FUNCTIONALITY	ENABLED
TFT display
QRCODE
ESPNOW	❌
PID
ETHERNET
CONVERT	❌
MQTT	❌
OLED
APDS9960
BME280	❌
INFRARED
PLAY (WAV, MP3, VOICE)
NEOPIXEL
VL53L0X
RFID
FREQUENCY METER
IOBUFFER	❌
WDT	❌
MPU9250
MPU6886
MPU6050
FUSION ALGO
BNO055
PCA9685
TM1637 AND TM1638
MAXDISPLAY

[PeterN]

Thanks  Francesco,

I do not see PWM in the table. Is it OK?

[cicciocb]

Hi Peter,

all that is not listed is included by default.

This list could evolve in the future depending on the requests.

[PeterN]

Thanks again, 

I have a X/Y-servo-mount for the ESPCAM  an I plan to try to follow a face with it.

[PeterN]

Hi CiccioCB,

I use face recognition  with an M5CAM. That works fine. You can enroll and recognize  7 faces, but the faces are stored in a kind of  ring buffer memory. So you can not exchange a specific single "old" face at a specific selectable memory position, but only add a new face and thus delete the eldest.  

Do you see a chance to add that kind of  function?

This is what I use at the moment for testing the face recognition:

CODE: xxxx.bas

---

' ##################################################################################
' M5Camera (or ESP32-CAM)  for face recognition
' Saves the last 7  detected faces in /faces.fh
' Ccompare the currently detected face with the stored fraces
'
'###################################################################################

rate = 30

rate = 10
detection_msg$ = "----"

F$ = "--"

'The Camera can recognize max 7 faceIDs and stored them in /faces.fh
dim FACE$(7)="PeterN","MonikaN","Obama","Johnson","Trump","Merkel","Face #7"
'dim FACE$(7)="Face #1","Face #2","Face #3","Face #4","Face #5","Face #6","Face #7"

' must set the resolution to 320*240 for facerecognition
print camera.setup(4) '320*240
print camera.params("vflip", 1)
print camera.params("hmirror", 1)

print camera.params("detection",1)
print camera.params("readfaces",1)
print camera.params("recognition",1)

'################################################
OnCamDetect FACE_DETECTED
onHtmlReload setpage
gosub setpage

wait

'###################################################################################
'###################################################################################

setpage:
'----------
cls
a$ = ""
A$ = A$ + |<h2> AnnexCAM Face Recognition</h2>|
A$ = A$ + |<img id='camera' src="picture" alt="AnnexCam" style="width:auto;"><br><br>|
A$ = A$ + "<br><br>"

A$ = A$ + textbox$(F$)

A$ = A$ + "<br><br>"
A$ = A$ + button$("V Flip", v_flip) + button$("H mirror", h_mirror)
A$ = A$ + "<br><br>"
A$ = A$ + button$("ENROLL THE NEXT FACE", ENROLL_FACES) + button$("Save all Faces to /faces.fh", SAVE_FACES)

OnCamDetect OFF

' Temporarily deactivate the face detection to prevent getting to many detection-events
'print camera.params("detection",0)

'The detected face may be one of the 7 faces stored in /face.fh

D$ = word$(detection_msg$, 2, "Match ID:")
if D$ <> "" then
  F$= FACE$(val(D$))
else
  F$= "?????"
end if
refresh
pause 1000
F$= "-"
'reactivate face detection
OnCamDetect FACE_DETECTED
'print camera.params("detection",1)
return

[cicciocb]

I think that it could be possible but I must check how do it.

I'll let you know.

[Florian Bednarz]

HI,

i have a 3d printed robot with 2x endless rotation Servos as drives

whenever i want to use PWM.setup the program stops or the esp32-cam is not reachable any more..

here my example code:

'title$ = "Virtual Joystick"

joy$ = "C"

jsexternal "/joy_cb.min.js"

rate = 30 ' rate between each picture sample (msec)

'pause 100

' set the max resolution in terms of memory

if camera.setup(10) = 0 then end

print camera.params("vflip", 1)

print camera.params("hmirror", 1)

'set at 800x600

print camera.params("framesize", 4)

speed = 20

leftservo =295

rightservo = 307

leftcenter = 300

rightcenter =307

PWM.SETUP 13, 1, leftcenter, 50, 12

PWM.SETUP 12, 2, rightcenter, 50, 12

onHtmlReload setpage

gosub setpage

'onHtmlChange input_check

wait

 

setpage:

cls

a$ = ""

a$ = a$ + |<h2>AnnexCAM Demo mini</h2> |

a$ = a$ + |<img id='camera' src="picture" style=width:25%;height:auto;">|

a$ = a$ +  "<br><br>"

'a$=a$ + |<div style='display: table; margin-right: auto; margin-left: auto;'>| + textbox$(title$,"t") + |</div>|+"<br>"

a$=a$ + |<div style='display: table; margin-right: auto; margin-left: auto;'>|

a$=a$ + |<div id="joyDiv" style="width:200px;height:200px;margin-bottom:20px;margin:10px"></div>|

a$=a$ + |<br><div style="display: table; margin-right: auto; margin-left: auto;">| + textbox$(joy$,"t") + |</div>|

a$=a$ + cssid$("t","text-align:center;border-style: none;")

autorefresh 100

html a$

a$ = |joy = new JoyStick('joyDiv');|

a$ = a$ + |joy.SetCallback(function(e) {connection.send('cmd:immediatxJOY$="' + e.GetDir() + '":HtmlEventVar$="joy$"\n' )});|

jscript a$

a$ = ""

jscall "set_pictimer(" + str$(rate) + ");"

return 'setpage

input_check:

if joy$ = "N" then

PWM.out 1, leftcenter - speed

pwm.out 2, rightcenter + speed

end if 'Nord                           'joystick activity returns directions directly into joy$ textbox

if joy$ = "S" then

PWM.out 1, leftcenter + speed

pwm.out 2, rightcenter - speed

end if

if joy$ = "E" then

PWM.out 1, leftcenter - speed

pwm.out 2, rightcenter - speed

end if

if joy$ = "W" then

PWM.out 1, leftcenter + speed

pwm.out 2, rightcenter + speed

end if

if joy$ = "C" then

PWM.out 1, leftcenter

pwm.out 2, rightcenter

end if 'Nord

return 'input_check

[ciccio cb]

The pins 12 and 13 are not available if you are using the SDCARD. 

You should put SDCARD NONE in the config page if you want to use these pins. 

--
You received this message because you are subscribed to the Google Groups "Annex WiFi RDS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to annex_wifi_rd...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/annex_wifi_rds/13ccd9f9-0582-475a-b53e-e8cb2dd0f813o%40googlegroups.com.

[Florian Bednarz]

> To unsubscribe from this group and stop receiving emails from it, send an email to annex_w...@googlegroups.com.

>
> To view this discussion on the web, visit https://groups.google.com/d/msgid/annex_wifi_rds/13ccd9f9-0582-475a-b53e-e8cb2dd0f813o%40googlegroups.com.

I know, so i disabled the sdcard in config...

Dos still not work...

[cicciocb]

I remember I did that trick (the double servo on the CAM) some time ago but  I used the GPIO12 and the RX pin.

Probably the SPI bus is always initialised in the Annex version that you have.

[Florian Bednarz]

Thanks a lot, i will try differnt pins.

i only used that two pins because the original arduino sketch (and so my add on pcb) used these two pins...

this is the thing i have printed:

https://www.thingiverse.com/thing:3987784

but i don´t like the original interface, so i want to build my own with annex basic ;-)

Flo

[Florian Bednarz]

I tried with many different pins: 01,03,12,13

whenever i do 

PWM.SETUP PIN, 1, 150, 50, 12

as mentioned in the Online help 

Annex stops working.

i tried 3 differnt esp32-cam modules from different Vendors...

always the same thing:

Annex output disconnects, no picture is shown and i have to restart the module manually to reach the web editor again...

i tried with every combination in the config screen, enable SD, disable SD etc...

this is my sample code:

Code hier eingeben...rate = 30 ' rate between each picture sample (msec)

pause 100

'when i include this line, no matter what PIN annex-cam stops working...
PWM.SETUP 12, 1, 150, 50, 12

' set the max resolution in terms of memory

if camera.setup(10) = 0 then end

print camera.params("vflip", 0)

print camera.params("hmirror", 0)

'set at 800x600

print camera.params("framesize", 6)

pause 100

onHtmlReload setpage

gosub setpage

wait

setpage:

cls

a$ = ""

'a$ = a$ + |<h2>AnnexCAM Demo mini</h2> |

a$ = a$ + |<br><br> |

a$ = a$ + |<img id='camera' src="picture" style="width:auto;height:25%;transform:rotate(90deg);">|

html a$

[Florian Bednarz]

Found a solution!

if i use PWM channel 3 and 4 instead of 1 and 2 it works perfect!

PWM.SETUP PIN, 3, 150, 50, 12