Amb82 Mini Arduino Framework Face Recognization Demo GUI With DumbDisplay

In this post, I will demonstrate my experiment on face recognition with the AMB82 MINI microcontroller board. The microcontroller program is developed with the Arduino framework using Arduino IDE.

The sketch is modified from one of the examples that come with the board installation to the Arduino IDE -- Examples/AmebaNN/RTSPFaceRecognition

What was added is simply some simple GUI for that example with the help of DumbDisplay; the spotlight stays at face recognition driven by AMB82-MINI.

Next to AMB82-MINI's capability of face recognition, its capability to stream video shot with its camera in RTSP is what makes the GUI look more appealing.

As a result, the GUI is simply comprised of:

1. a way to show the RTSP stream
2. a way to ask for user input for the name of the person whose face is to be registered
3. a way to highlight the registered name when the person's face is recognized

The sketch of this experiment is actually bundled with the installation of the DumbDisplay Arduino library; hence, it is easy to get hold of the sketch, as will be described next.

The steps to install AMB82-MINI to your Arduino IDE are:

1. open preference dialog -- "File" -> "Preferences"
2. enter to "Additional boards manager URLs" -- https://github.com/ambiot/ambpro2_arduino/raw/main/Arduino_package/package_realtek_amebapro2_index.json
3. select board -- "Tools" -> "Board" -> "Boardd Manager"
4. in Board Manager, search for "ameba" and install the board "Realtek Ameba Boards (32-bit Arm v8M @ 500MHx)"

After a moment, you should be able to select AMB82-MINI microcontroller board as the target of your sketch development

To be sure that the board is installed correctly, you might want to first try out the standard blink example [of the Arduino IDE]:

1. "File" -> "Examples" -> "01. Basic" -> "Blink"
2. upload the sketch
3. connect your computer to your board -- with the Micro USB (CH340) USB socket
4. select board -- "Tools" -> "Board" -> “Realtek Ameba Boards (32-bit Arm v8M @ 500MHx)” -> “AMB82-MINI”
5. select the port you connected your board to your computer
6. put your board in "upload" mode -- depress the "UART_UPLOAD" button then click the "RESET" button (as illustrated with an animated GIF above)
7. upload the sketch
8. reset your board by clicking the "RESET" button

See that the blue LED on your board is blinking, as shown with another animated GIF above

The steps to install the DumbDisplay Arduino library to your Arduino IDE are as easy:

1. "Tools" -> "Manage Libraries" -- search "dumbdisplay"
2. install the library "DumbDisplay by Trevor Lee"

Now, DumbDisplay Arduino library is ready to be used in your sketch programming

As mentioned previously, the sketch of this experiment -- amb82_facerecog -- comes bundled with the DumbDisplay Arduino library:

1. open the amb82_facerecog example -- "File" -> "Examples" -> "DumbDisplay" -> "amb82_facerecog"
2. save a copy, say with name like "my_amb82_facerecog"
3. since AMB82-MINI will need to access WiFi for RTSP, you will need to provide your WiFi access credentials by adding _secret.h
4. create new file "_secret.h" -- click "New Tab" of the tab bar
5. put to the file

#define WIFI_SSID "wifi ssid"

#define WIFI_PASSWORD "wifi password"

1. upload the sketch
2. connect your computer to your board -- with the Micro USB (CH340) USB socket
3. select board -- "Tools" -> "Board" -> “Realtek Ameba Boards (32-bit Arm v8M @ 500MHx)” -> “AMB82-MINI”
4. select the port you connected your board to your computer
5. put your board in "upload" mode -- depress the "UART_UPLOAD" button then click the "RESET" button
6. upload the sketch
7. reset your board by clicking the "RESET" button

See it working with your Android phone:

1. unplug the USB cable connected to your computer, and plug the cable to your Andriod phone (via OTG adapter) -- you may want to refer to my previous post Blink Test With Virtual Display, DumbDisplay
2. open DumbDisplay Android app
3. from DumbDisplay Android app, make USB connection to your AMB82-MINI board (orange item as shown in the above screenshot)

Note that the AMB82-MINI face recognition processing outputs a considerable amount of logging info to the Serial port, which is also the channel used by DumbDisplay Arduino library to communicate with DumbDisplay Android app; as a result, such communication "noise" might sometimes mess things up (resulting in unexpected GUI interactions), that you unluckily have to live with :-(

The GUI realized with DumbDisplay Android app is in fact driven by the sketch running on your AMB82-MINI board, and is simple:

1. you click the "Register" button to register the face showing by giving the face a name
2. once registered, the face should be recognized with the name highlighted
3. at this point, the "💾" button should be enabled for you to save your registered face(s)
4. in case you want to remove any registered face(s), first stop video streaming by clicking "📺"
5. then you can select the name [of a face] to deregister
6. when done, click "📺" again to enable face recognition again

Note that if you save your registered faces by clicking"💾" button, the faces will be "remembered" even you reboot your AMB82-MINI board

Hope that you will have fun with it! Enjoy!

Peace be with you! May God bless you! Jesus loves you! Amazing Grace!