Creating socially awkward wearables with ML & JS
summary
A wearable necklace uses a camera on a choker to snap photos every few seconds, runs inference on a Raspberry Pi running Node.js and TensorFlow.js to count nearby people, and sends a single byte over serial to a microcontroller that drives LEDs. Behavior changes with thresholds (calms when count ≥ six), includes a manual long‑press kill switch, and documents debugging wins: swapping controllers for voltage stability, eliminating a memory leak, and replacing bulky JSON with compact binary messages.
AI-written summary · about
chapters (21)
- 0:00Intro
- 1:04About Stephanie
- 3:22The whole outfit
- 4:44Why this project
- 7:56Why I made it
- 9:19Why I built it
- 10:26Resources
- 12:38Models
- 14:20Raspberry Pi
- 15:30Disclaimer
- 16:13Reusing Hardware
- 18:41Soldering
- 19:28Serial
- 21:09Power Issues
- 21:44Switching Boards
- 22:33Memory Leak
- 23:09Debugging
- 26:51Adrenal
- 28:22Problems
- 29:50Button Ring
- 31:01Conclusion
related talks
26mAlice in Hardware Land - JavaScript models of interaction with the physical world
14mFrom JS to manufacturing, a hardware journey
18mExploring alternative interactions in JavaScript
44mIntroduction to JavaScript Robotics
42mExploring Alternative Interactions in JavaScript
27mHow I've been using JavaScript to automate my house
26mFrom Software to Hard ware: How Do I Track My Cat with JavaScript
42mJavaScript in All the Creative Places