From Data to AI on The Edge
summary
Trains a custom image classifier with Azure Machine Learning, exports it as a container, and deploys it to devices via Azure IoT Edge; demo shows a Raspberry Pi running the container that classifies camera frames locally and flips lights as an alarm with low latency. Uses Custom Vision for labeling, exports Docker containers and a deployment manifest, and manages device rollouts from the cloud; the end-to-end pipeline works live on a trash-defender prototype. Azure Machine Learning Azure IoT Edge Raspberry Pi
AI-written summary · about
chapters (22)
- 0:00Welcome and introduction
- 1:05What is machine learning
- 2:11Trash defense system demo
- 3:33How machine learning works
- 5:21Azure AI services overview
- 7:08Custom Vision service demo
- 10:39Creating a new project
- 13:12Training and testing models
- 16:06Adding more training images
- 17:37Object detection projects
- 20:09Exporting models to devices
- 22:42Real-time trash monitoring
- 24:49Edge AI advantages
- 26:35Cloud and edge integration
- 27:15Azure IoT Edge overview
- 30:53Live system demonstration
- 33:26Connecting devices to Azure
- 35:57Edge project structure
- 38:27Updating models on devices
- 40:59Monitoring and testing
- 42:27Edge AI use cases
- 43:11Additional resources
related talks
01h 02mAI on the Edge
49mEnable IoT with Edge Computing and Machine Learning
43mIntersection of Awesome: Building AI enabled IoT solutions powered by the cloud
06mLightning Talk: Deploying AI Solutions on the Edge with Containers
37mPracticing Data Science in the cloud
31mBuild Your Intelligent Edge
44mGetting started with Azure Machine Learning services
01h 00mLearn Live: Get started with AI on Azure