GABC2019: Sort your (child's) Legos with Cognitive Services by David Schneider

With Custom Vision we can recognize and assign objects. A simple example teaches you how to tag Lego bricks, train the model, and call the model from your code. With Custom Vision you can sort the Lego bricks by color, size or type. In this session you will learn the power of Custom Vision in a simple and fun way and you will see how to approach such a project.

Transcript

1. David Schneider: Sort your (child's) Legos with Cognitive Services
2. About me David Schneider CTO & Office Manager Barcelona https://sharepointcommunity.ch https://linkedin.com/in/david-schneider https://twitter.com/fiddi https://davidschneider.blog
3. Thanks to our sponsors!
6. Pre-Built AI 6 Ready to use with API For a reasanable price Performant Customizable Compliant In the Cloud or on the edge
7. COGNITIVE SERVICES
9. SCENARIOS Wear & Rust Detection Incorrect Configurations Sort Parts
10. CUSTOM VISION Upload Images Upload your own labeled images Train Teach Custom Vision services using your labeled images. Evaluate Use REST API to tag images with your custom vision model.
11. OBJECT DETECTION VS. CLASSIFICATION blue-4x2: 99% green-4x2: 2% green-4x2: 98% blue-4x2: 2%
12. Demo: Custom Vision
13. Custom Vision SDK
14. COMMONLY USED APIS Prediction Training • Create images • Tag images • Create projects • Manage projects • Manage iterations • Create tags • Get account information • Train a project • Predict images • Predict and save images • Predict image URLs • Predict and save image URLs
15. Demo: Prediction in C#
16. Custom Vision & Azure IoT Edge
17. TRAIN IN THE CLOUD, RUN ANYWHERE Train your model in Custom Vision Service Deploy & Run Anywhere TensorFlow ONNX/WinML CoreML Azure IoT Edge Docker Export
18. AZURE IOT EDGE + CUSTOM VISION Camera Custom Vision Display Azure IoT Edge Runtime IoT Hub & Container Registry Camera https://github.com/Azure-Samples/Custom-vision-service-iot-edge-raspberry-pi
19. Demo: Custom Vision Model on a Raspberry Pi
20. VISION AI DEV KIT https://visionaidevkit.com
22. Thank you
23. Thanks to our sponsors!