Lesson 3: AI-Powered Smart Glasses (OpenGlass Pt. 1)
In this lesson, you’ll transform your XIAO ESP32 S3 Sense into AI-powered smart glasses using the OpenGlass open-source project. This is your first major step into wearable, real-time computer vision—where your device can “see” text in the world and extract it instantly using onboard OCR (optical character recognition).
🎯 What You’ll Accomplish Today:
- Install and run the OpenGlass code on your ESP32
- Use the onboard camera to extract and display live text data
- View that data in real time via the Arduino IDE Serial Monitor
- Begin exploring how low-power AI devices can sense and interpret their environment
🔍 What Is OpenGlass?
OpenGlass is an open-source project designed to:
- Use the ESP32’s camera to capture images
- Run OCR to extract readable text
- Output that data through serial communication
- (Optionally) integrate with AI tools in future lessons
You’re not modifying or extending the project today—just getting it installed, running, and tested successfully on your own device.
📎 GitHub repo: https://github.com/BasedHardware/OpenGlass
🛠️ Setup Instructions:
Step 1 – Download and Open the Code
- Download the OpenGlass project as a ZIP
- Unzip it and open the Arduino sketch in your IDE
- Set your board to XIAO_ESP32S3 and select the correct COM port
Step 2 – Board Package Compatibility
⚠️ OpenGlass may not compile correctly on the latest ESP32 board versions.
- Go to Tools > Board > Board Manager
- Search for "ESP32" by Espressif Systems
- Downgrade to version 2.0.17 for best compatibility
- Install any missing libraries listed in the repo’s README
Once ready, hit Upload to flash the code to your board.
👓 Step 3 – Test the OCR Smart Glasses
- Open the Serial Monitor in Arduino IDE
- Hold up printed or digital text in front of the ESP32’s camera
- Watch as the OCR engine extracts the text and prints it to the monitor in real time
Try scanning:
- Book pages
- Text on your phone screen
- Labels, signs, or printouts
This is a completely local system—no cloud, no Wi-Fi required—just smart wearable vision.
🧪 Hands-On Activity:
- Test the glasses on three types of text:
- Printed book
- Screen display
- Handwritten note
- Note which conditions (lighting, distance, font) give the best results
🧾 The video includes side-by-side comparisons of clean vs. blurry reads to help you calibrate.
🛠️ Troubleshooting Tips:
If you’re not seeing readable text:
- Make sure the camera is not blocked or misaligned
- Ensure the Serial Monitor is set to the correct baud rate (check the repo)
- Re-upload the code if you saw compile errors
- Confirm you're on ESP32 board version 2.0.7 or 2.0.9
📚 Homework Assignment:
- Capture at least 5 readable OCR results
- Post one interesting or funny result in the #ai-glasses-log Discord thread
- (Optional) Jot down two ideas for how you might extend this project (e.g., speech output, text-triggered alerts)
✅ By the End of This Lesson, You’ll Have:
- Installed and run the OpenGlass OCR engine
- Converted your ESP32 glasses into a real-time text-reading AI device
- Started thinking about real-world applications for low-power computer vision
In Lesson 4, we’ll explore how to connect this visual data to external actions—like triggering smart home automations or sending alerts.