Build a custom Edge Machine Learning solution to monitor your shipments conditions with Edge Impulse
summary
Creates a full Edge ML pipeline that classifies package motion and deploys to a Generic Node: data captured on a phone, an impulse with spectral DSP plus a Keras classifier trained with an 80/20 split, sampling set to 50 Hz, a small dense network (layers of 20 and 10 neurons with dropout), model quantized and optimized with the EON compiler, exported as a C++ library, imported into STM32CubeIDE, flashed to the board and run — the compiled firmware performs inference and correctly labels a held sample as driving.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:13Create a new project
- 4:27Create Impulse
- 5:30Feature Explorer
- 6:23Neural Network Classifier
- 7:35Deployment
- 12:40Testing
related talks
05mBuilding Machine Learning models with Edge Impulse
45mUsing Edge Machine Learning to save power and development time
49mEnable IoT with Edge Computing and Machine Learning
15mSimplifying & Scaling Edge AI
20mScaling Edge AI with Elixir and Nerves
30mTurn Any Android Devices into an AI Sensor Hub
45mEdge Impulse & Android: Democratizing AI at the Edge
31mBuild Your Intelligent Edge