Capturing a Day in Life with Smart Glasses for AI Analysis
The author built a smart glasses recording system to capture a full day of real-world activities and pass the data to AI for analysis. The system uses Rokid Glasses, a smartphone, and a backend to process the footage and make it browsable and queryable.
Why it matters
This experiment demonstrates how wearable devices can capture real-world context to enhance AI systems, while also highlighting the technical and privacy challenges that need to be addressed.
Key Points
- 1Captured 13 hours of daily activities using smart glasses
- 2Addressed practical challenges like storage and battery life
- 3Prioritized privacy with easy recording on/off and indicators
- 4Developed a web-based UI to browse the captured timeline
- 5Used multimodal AI to index and make the data searchable
Details
The author wanted to explore giving AI more personal context beyond just digital data like journals and screen recordings. Smart glasses were chosen as the capture device due to their form factor, camera, microphones, and ability to run an Android-based system. The system splits recordings into chunks and transfers them to a smartphone for storage and relay to a backend for processing. A neck-mounted power bank ensures all-day battery life. Privacy was addressed with easy recording controls and indicators. The captured footage was made browsable through a web-based timeline UI, and then indexed using multimodal AI to enable searching and querying the data.
No comments yet
Be the first to comment