Greetify

AI-Based Greeting System with XIAO ESP32S3 Sense

A Smart Interactive Signage that Detects People and Displays a Scrolling Welcome Message

📖 Introduction

In this project, we will build an AI-powered greeting system that detects when a person is nearby and displays a scrolling welcome message on a 16×2 LCD screen.

We’ll use the Seeed Studio XIAO ESP32S3 Sense, a powerful microcontroller with a built-in camera and microphone, and leverage SenseCraft AI to run a TinyML person detection model locally — no internet connection required!

This project is perfect for beginners who want to explore AI, TinyML, and IoT hardware while creating something fun and practical for offices, exhibitions, events, or even your home.

🎯 Features

• AI-based Person Detection using a pretrained TinyML model
• Scrolling LCD Message: “Welcome to Stark Tower” when a person is detected
• Blank Screen when no one is present
• Runs fully offline on a small board
• Beginner-friendly: No prior AI experience needed

🛠️ Hardware Required

Seeed Studio XIAO ESP32S3 Sense

16×2 I²C LCD Display

Breadboard

Jumper Wires

USB-C Cable

🖼️ How It Works

SenseCraft AI Model: A pretrained person detection model runs on the XIAO ESP32S3 Sense camera module.

AI Processing: The board continuously checks for people in view.

LCD Output:

If a person is detected → Scroll “Welcome to Stark Tower” across the screen.

• If no person is detected → Keep the screen blank.

📋 Step-by-Step Guide

Step 1 — Set Up Arduino IDE

Install Arduino IDE.

• Install Arduino IDE.

In File → Preferences → Additional Board Manager URLs, add:

https://espressif.github.io/arduino-esp32/package_esp32_index.json

Go to Tools → Board → Board Manager, search for esp32, and install the latest version.

Select the board:Tools → Board → ESP32 → XIAO_ESP32S3.

Step 2 — Set Up SenseCraft AI

We’ll use SenseCraft AI from Seeed Studio to easily run pretrained ML models.

Follow these guides:

Getting Started with XIAO ESP32S3 Sense

Deploy Pretrained Models

SSCMACore Library

In summary:

Download and install the SenseCraft AI Arduino library.

From the SenseCraft AI website, download the Person Detection pretrained model.

Deploy it to your XIAO ESP32S3 Sense via the Arduino IDE or SenseCraft uploader.

Step 3 — Wire Up the LCD

The LCD we’re using communicates over I²C, which means only 4 wires.

unsure of your LCD’s I²C address, run an I²C scanner sketch first.)

Step 4 — Install Libraries

In Arduino IDE, install:

LiquidCrystal_I2C (by Frank de Brabander or similar)

Seeed_Arduino_SSCMA (from Seeed Studio)

Step 5 — The Code

#include <Wire.h>
#include "Seeed_Arduino_SSCMA.h"
#include <LiquidCrystal_I2C.h>

SSCMA AI;
LiquidCrystal_I2C lcd(0x27, 16, 2); // Change 0x27 to your LCD I2C address

bool lastState = false;

void setup() {
    Serial.begin(115200);
    lcd.init();
    lcd.backlight();
    lcd.clear();

    if (!AI.begin()) {
        Serial.println("AI Init Failed");
        while (1);
    }
    Serial.println("AI Ready");
}

void loop() {
    SSCMA_Result result;
    if (AI.invoke(result)) {
        bool personDetected = false;

        for (int i = 0; i < result.count; i++) {
            if (String(result.objects[i].name) == "person") {
                personDetected = true;
                break;
            }
        }

        if (personDetected) {
            if (!lastState) {
                lcd.clear();
                lcd.setCursor(0, 0);
                lcd.print("Welcome to Stark Tower");

                for (int pos = 0; pos < 16; pos++) {
                    lcd.scrollDisplayLeft();
                    delay(300);
                }
            }
        } else {
            if (lastState) {
                lcd.clear();
            }
        }
        lastState = personDetected;
    }
}

Step 6 — Upload and Test

Connect your XIAO ESP32S3 Sense via USB-C.

Select the correct port in Arduino IDE.

Click Upload.

💡 Future plans

Multi-language greetings (e.g., English in the morning, local language in the evening)

LEDs or buzzer to add audio/visual effects

Integration with Wi-Fi to log visitor counts online

Name recognition using custom-trained ML model

🎥 Demo Video

🏆 Why This Fits the Seeed Interactive Signage Contest

This project is:

Interactive — responds instantly to people.

AI-powered — runs TinyML locally.

Beginner-friendly — can be repliated in under 1 hour.

Scalable — can be adapted for events, shops, or homes.

• Scalable — can be adapted for events, shops, or homes.