See More, Be More
summary
A native iPad app captures a depth map and guided head-movement video, sends image IDs to blob storage, and runs an asynchronous pipeline of in-house classifiers (face shape, eye color, palettes) plus migrated landmarking via MediaPipe and managed ML services (Microsoft) to score matches against a product catalog of 700,000+ frames. A Python orchestration function assembles parallel model outputs, returns recommendations and a virtual try-on in seconds, uses time‑bound tokens to hand off to Smart Shopper, and runs at retail scale across 18,000 stores.
AI-written summary · about
chapters (9)
- 0:00Introduction and company overview
- 7:51Business challenge and solution approach
- 9:16Frame advisor experience explained
- 13:54Face feature extraction process
- 17:09Classification models and architecture
- 21:06Product feature extraction and matching
- 25:46Smart Shopper and virtual try-on
- 27:32Live demo of the application
- 32:05Conclusion and talent recruitment
related talks
36mApps That See 👀 Practical AI Vision for Developers
32mAdvanced on-device ML made easy
33mComputer Vision, Not Just for Breakfast Anymore
22mEverything a Swift Dev Needs to Know About Machine Learning
32mSelf-Portraits in JavaScript
46mBuild apps with state-of-the-art computer vision
50mBuilding Intelligent Mobile Apps using Machine Learning
25mPersonalizing Your Images with AI-Powered Features