Skip to main content
This page collects example projects to help you start building with Proton AI Core as quickly as possible. Each example is self-contained and designed to demonstrate a key capability of the board — from live camera streaming to on-device AI inference. Pick the one closest to your use case and use it as a starting point.

Camera Web Server

Stream live camera footage from Proton AI Core over your local Wi-Fi network and view it in any browser.

Wi-Fi Image Upload

Capture a JPEG frame from the camera and POST it directly to a remote HTTP server endpoint.

BLE Sensor Node

Read sensor data and broadcast it to nearby BLE clients using notifications — no Wi-Fi required.

Person Detection

Run a TensorFlow Lite Micro person detection model on live camera frames and report confidence scores over Serial.

​
Example 1: Camera Web Server

The Camera Web Server example streams a live MJPEG feed from the OV2640 or OV3660 camera over Wi-Fi. Any device on the same network can view the stream by navigating to the board’s IP address in a web browser — no app installation required. Uses: esp32-camera, WiFi.h The Arduino IDE ships with a ready-made version of this sketch. You can find it at File → Examples → ESP32 → Camera → CameraWebServer.
1

Install the ESP32 board package

In Arduino IDE, open Tools → Board → Boards Manager, search for esp32 by Espressif Systems, and install it. The esp32-camera driver is bundled automatically.
2

Open the CameraWebServer example

Navigate to File → Examples → ESP32 → Camera → CameraWebServer and open the sketch.
3

Configure your Wi-Fi credentials

In the sketch, replace the placeholder values for ssid and password with your network name and password.
4

Select the correct camera model

Near the top of the sketch, uncomment the #define that matches your camera module (OV2640 or OV3660). Comment out all other camera model defines.
5

Upload and connect

Select ESP32S3 Dev Module (or the equivalent Proton AI Core board entry) under Tools → Board, then click Upload. Open the Serial Monitor at 115200 baud and wait for the board to print its IP address. Navigate to that address in your browser to view the stream.
Use FRAMESIZE_SVGA (800×600) in the web server sketch for a good balance of image quality and streaming frame rate over a local Wi-Fi network.

​
Example 2: Wi-Fi Image Upload

This example captures a single JPEG frame from the camera and sends it to a remote HTTP server using an HTTP POST request. It is useful for remote monitoring pipelines, cloud-based AI processing, or any workflow where you need to move images off the device. Uses: esp32-camera, WiFi.h, HTTPClient.h Connect to Wi-Fi and initialise the camera first (see Wi-Fi & Bluetooth and Camera), then call the function below:
wifi_image_upload.cpp
#include <WiFi.h>
#include <HTTPClient.h>
#include "esp_camera.h"

// After Wi-Fi connected and camera initialized...
void uploadImage() {
  camera_fb_t *fb = esp_camera_fb_get();
  if (!fb) return;

  HTTPClient http;
  http.begin("http://your-server.com/upload");
  http.addHeader("Content-Type", "image/jpeg");
  int code = http.POST(fb->buf, fb->len);
  Serial.printf("Upload response: %d\n", code);
  http.end();
  esp_camera_fb_return(fb);
}
Replace http://your-server.com/upload with your actual server endpoint. Your server should accept a POST request with a Content-Type: image/jpeg body and respond with an HTTP 200 on success.

​
Example 3: BLE Sensor Node

The BLE Sensor Node example turns Proton AI Core into a Bluetooth Low Energy peripheral that broadcasts sensor readings to any nearby BLE central device — a phone, tablet, or another microcontroller. No Wi-Fi infrastructure is needed, making it ideal for low-power or offline deployments. Uses: BLEDevice.h, BLEServer.h Start with the BLE Server sketch from the Wi-Fi & Bluetooth guide. To turn it into a sensor node, replace the static "Hello BLE" string in pCharacteristic->setValue() with your actual sensor reading. For example, if you are reading a temperature sensor connected over I2C, format the reading as a string or pack it as raw bytes before calling setValue(), then call notify() to push the update to subscribed clients. Use a BLE scanner app such as nRF Connect to verify that your characteristic value updates correctly before integrating with a full client application.

​
Example 4: Person Detection

The Person Detection example runs a pre-trained TensorFlow Lite Micro model on live camera frames and reports a confidence score for “person” and “no person” directly to the Serial Monitor. It demonstrates the full edge AI pipeline: camera capture → pre-processing → on-device inference → result output. Uses: esp32-camera, TFLite Micro, WiFi.h (optional, for OTA updates) Espressif maintains a reference implementation in their tflite-micro-esp-examples repository, which includes a person detection demo optimised for the ESP32-S3.
1

Clone the repository

git clone --recursive https://github.com/espressif/tflite-micro-esp-examples.git
2

Open the person detection example

Navigate to the examples/person_detection directory inside the cloned repository.
3

Configure PlatformIO

Open the project in VS Code with the PlatformIO extension installed. Update platformio.ini if needed to target the ESP32-S3 and set the correct upload port.
4

Flash and observe

Build and upload the firmware. Open the Serial Monitor at 115200 baud. Point the camera at a person or an empty scene and observe the “person” and “no person” confidence scores printed for each inference cycle.
Full Proton AI Core-specific example code is coming soon. In the meantime, the Espressif reference examples run on the ESP32-S3 and serve as an accurate starting point.