AI in the battle against fakes
summary
A solution identifies counterfeit electronics from a single smartphone photo by combining object detection (icon/logo cropping) with a logo classifier exported as a container and served at scale; prototype labels use ~50 images per class (demo model reached near 100% on held examples but needed more data in production). Architecture uses Azure Functions, Cosmos DB, Docker for orchestration and storage and Azure Machine Learning service, Custom Vision for training, export and CI/CD via containers.
AI-written summary · about
chapters (8)
- 0:00Introduction to AI against fakes
- 5:09Project overview and goals
- 11:34Item detection and validation demo
- 23:23Custom Vision service walkthrough
- 27:42Machine Learning service setup
- 35:48Training models in notebooks
- 41:51Testing and deploying models
- 47:48Q&A and closing remarks
related talks
59mAI in the battle against fakes
59mAI in the Battle Against Fakes
01h 02mAI on the Edge
01h 04mBuilding and Training your own Custom Image Recognition AI
42mProduct Recognition in Retail/E-Commerce with Azure AI
42mProduct Recognition in Retail/E-Commerce with Azure AI
45mFrom Data to AI on The Edge
41mAI + Machine Learning for .NET Developers