LifeX – The All-in-One Smart Companion on STM32N6570-DK

LifeX is an all-in-one smart companion built on the STM32N6570-DK, combining security, monitoring, fun, and music into a single system. From detecting wildlife at night, tracking personnel during the day, snapping creative photobooth pictures, to sensing your dance moves and playing matching music — LifeX transforms a development board into a versatile, interactive, and intelligent daily companion.

LifeX – The All-in-One Smart Companion on STM32N6570-DK

Hi! In this tutorial, I’ll walk you through my project LifeX – a multi-application smart system built on the STM32N6570-DK board using TouchGFX.

I designed LifeX to bring together serious security features, useful monitoring tools, and a touch of fun and creativity – all inside one unified system.

Why I Built LifeX

I wanted my STM32N6 board to be more than just a demo platform. Instead of one single-purpose app, why not make it a multi-purpose system that adapts to different times of the day and user needs?

At night, it should keep me safe.
During the day, it should help with personnel tracking.
In free time, it should let me have fun with friends.
And when I want to relax, it should even play music based on my moves.

That’s how LifeX was born

Features in LifeX

LifeX packs four applications in one system:

Night Guardian
- Detects wild animals (like tigers) at night.
- Logs detection events with timestamps.
- Acts as a mini security system.
Day Vision
- Applies a heat-vision style filter to highlight people.
- Logs personnel faces for attendance/security.
SnapBooth 3X
- Takes 3 fun photos with filters.
- Generates a QR code to download the pictures from an ESP32 server.
GrooveSense
- Detects dance movements.
- Plays music that matches the style of movement.

All of this is managed by a wheel-style App Manager, so I can quickly switch apps.

How I Built LifeX

Hardware

STM32N6570-DK board
Camera module
ESP32 (for photobooth QR hosting)
Speaker/headphone output

Software

STM32CubeIDE
TouchGFX Designer
STM32CubeProgrammer

Project Structure

I organized the project into app-specific folders:

gui/
├── nightguardian_screen/ # Night Guardian App
├── dayvision_screen/ # Day Vision App
├── snapbooth_screen/ # Photobooth
└── groovesense_screen/ # Mood Music

Each app has its own assets, while the Model keeps everything connected.

Edge Impulse

Understanding the files

network_data.hex → Compiled weights for STM32 (X-CUBE-AI runtime).
- Hex is basically a binary dump you cannot directly include in C code.
What we need → network_data.c and network_data.h:
- .c contains the byte array of the model.
- .h contains function declarations and constants.

Edge Impulse usually generates this via their STM32 deployment/export.

Generate network_data.c and network_data.h

Open your project on Edge Impulse Studio.
Go to Deployment → STM32Cube.AI.
Click Build or Download for STM32Cube.AI.
You’ll get a ZIP containing something like:

network_data.c network_data.h network_initialize.c network_initialize.h

network_data.c → contains your model weights in a const unsigned char network_data[].
network_data.h → contains extern declaration and some helper macros.
The network will output 4 classes, based on the labels you set in Edge Impulse (boar, tiger, deer, elephant).

Simple Inference Code

float input[INPUT_SIZE]; // your preprocessed sensor/image data float output[OUTPUT_SIZE]; // 4 classes
ai_network_run(network, input, output);
// Map output to labels const char* labels[4] = {"boar", "tiger", "deer", "elephant"}; int max_idx = 0; for (int i=1; i<4; i++) { if (output[i] > output[max_idx]) max_idx = i; }
printf("Predicted: %s\n", labels[max_idx]);

Step-by-Step Build

Open the Project in STM32CubeIDE.
Load TouchGFX Designer and open contest_Project_hBz.touchgfx.
Generate code for the GUI.
Build the firmware inside CubeIDE.
Flash it to the STM32N6570-DK with CubeProgrammer.
Run the board – the App Manager appears first, then I can switch between apps.