Edge Computing and next-gen of IoT Sensors
summary
Prototype sensors fuse accelerometer gyro microphone and camera into waterproof battery nodes and run on-device ML models (Octo Nyan rhenium on Cortex-M class and GreenWaves GAP8 for deep nets). A tiny 80×64 camera on a Cortex-M4 takes 1 frame per minute; on GAP8: 30 fps the same people-count network duty-cycled to one detection every 10 minutes yields >10 years battery life versus ~2 months on M4. Work is demonstrator-level and not yet production ready.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:32The Machine Learning Frenzy
- 1:40Computing for the Internet of Things
- 2:36Typical Computing Platforms
- 4:54Developing machine learning applications
- 7:47Reducing energy consumption
- 9:26Miromico Ecosystem
- 9:47Artificial Intelligence
- 10:16Deep Learning in Embedded Systems
- 11:39GAPuino Board
- 12:31Energy efficient hardware for next-gen loT
- 14:14Conclusion
related talks
39mFrom a Weekend Hack, to a Future-Proof IoT Device
59mML and the IoT: Living on the Edge
31mKeynote: The next big thing in machine learning is tiny
22mAI On The Farthest Edge
39mFrom a Weekend Hack, to a Future-Proof IoT Device
54mSmarter Everything with ML & the IoT
35mData science and Internet of Things on the edge
16mRVEdge-Vision: A Fully Open,Ultra-Efficient On-Device AI Platform for Smart Eyewear