How not to read the room: Creating socially awkward wearables
summary
Build: a LED necklace with a camera on a Raspberry Pi 3 B+ running Node.js and TensorFlow.js that counts people with a tiny YOLO model classifies up to 20 objects and streams a single byte count over serial to an Adafruit Metro controlling blinky tiles. Key engineering facts: had to compile TensorFlow.js for ARM, solved a memory‑leak and power issues, swapped to a 5V board, and reduced messaging to a three‑byte serial protocol; the wearable works as intended.
AI-written summary · about
chapters (37)
- 0:00Intro
- 1:42Im fascinated with making myself uncomfortable
- 2:25Fiber optic dress
- 3:13Jellyfish dress
- 3:40Hallit Couture
- 5:22Speech to Image Necklace
- 6:38Machine Learning
- 7:07My current project
- 9:07Machine Learning for Artists
- 9:46Tensorflow
- 11:16Node
- 12:12Tweet
- 12:26Getting it running
- 12:53Building on the PI
- 13:26Disclaimer
- 14:13Blinky Tiles
- 14:47Micro Controller
- 15:10DMX Simple
- 15:39TTT Board
- 15:52Designing
- 16:14Cutting wires
- 16:24Serial ports
- 16:51Loopback
- 17:47Troubleshooting
- 18:11Power Issues
- 19:04Memory Leak
- 19:21Machine Learning Model
- 19:33Memory Leaks
- 20:07JSON
- 21:08Arduino
- 22:02Lights
- 22:20Code
- 23:27Hardware
- 24:00Straps
- 24:21Turning it off
- 24:38Button Ring
- 25:25Conclusion
related talks
25mHaute Codeture: Clothing as a Canvas
29mCreating Art with a Raspberry Pi
19mAdventures with blinking lights
02mRainbow of Raspberry
27mMore Fun With Hardware and CircuitPython - IoT, Wearables, and more!
46mInteractive Light
08mDemos Her Raspberry Pi Driven Wearables
14mFrom JS to manufacturing, a hardware journey