Building a Security Camera with AI using Cloudflare

Overview

This project demonstrates how to build a security camera system leveraging Cloudflare Workers, Workers AI, and R2 for real-time object detection and storage. By using an ESP32-CAM to capture images and Cloudflare’s edge infrastructure for AI processing, this solution is scalable, cost-effective, and highly performant.

This project was made for a IRL meetup at the Cloudflare Lisbon Office.

Features

ESP32-CAM Integration: Captures images every 5 seconds and sends them to a Cloudflare Worker.
Object Detection: Workers AI analyzes the images to detect objects, providing labels, bounding boxes, and confidence scores.
Cloudflare R2 Storage: Stores images in Base64 format and object detection results as JSON files.
Real-Time Web UI: Visualizes the results with red bounding boxes and labels for detected objects.

Final Result

Hardware:

Web UI

System Architecture

graph TD
    A[ESP32-CAM] -->|Captures Image| B[Cloudflare Worker]
    B -->|Sends to AI Model| C[Workers AI]
    C -->|Detects Objects| D[Cloudflare R2]
    D -->|Stores Images & Metadata| E[Real-Time Web UI]
    E -->|Fetches Data| D

Prerequisites

ESP32-CAM module
Cloudflare account with access to:
- Workers
- Workers AI
- R2 storage

Setup

1. Configure the ESP32-CAM

All of the esp32-cam code is from https://RandomNerdTutorials.com/esp32-cam-http-post-php-arduino/ I've just added a simple blink feature when the esp is taking a picture.

Full code is available in the esp32-uploader.ino

Update the code to point into your own worker and your WIFI ssid and password
Flash your ESP32-CAM using the instruction in the RandomNerdTutorials linked above

2. Deploy the Cloudflare Worker

Install dependencies and login with Cloudflare

npm install

npx wrangler login

Create a R2 Bucket

npx wrangler r2 bucket create esp32-cam-with-workers-ai

Deploy the worker

npx wrangler deploy

Usage

Power on the ESP32-CAM to start capturing images.
Monitor the Web UI for real-time results.
View stored images and detection metadata in R2.

License

This project is licensed under the MIT License.

G4brym/esp32-cam-with-workers-ai