Retail tracking edge2cloud system
summary
A small edge‑to‑cloud retail system is built for real‑time video emotion and face analytics, prototyped on a Coral Dev Board with TensorFlow and OpenCV and initially piped through Amazon Kinesis, Redshift, Lambda, SNS, Quicksight before simplifying to on‑device inference and SQL metric storage. A demonstrator UI runs on the device, processes live streams, and shows mood heatmaps; a 100‑store pilot in Ukraine installs devices in one or two shops to collect data. Age/gender and clustering models are work in progress.
AI-written summary · about
chapters (14)
- 0:00Intro
- 1:03Retail tracking Edge2cloud system
- 5:25Retail domain highlights
- 7:41Solution goals
- 10:23Business features
- 13:20Solution advantages
- 18:12Extendability
- 19:58Benefits?
- 20:18Solves business needs
- 21:09Implements innovations
- 22:40What's next?
- 24:00Further evolution
- 25:25Implementation vectors
- 29:31Demo
related talks
15mBuild a custom Edge Machine Learning solution to monitor your shipments conditions with Edge Impulse
12mWeb AI in industry: How TensorFlow.js has driven what you see on the supermarket shelves
57mWhat's that fruit?
01h 02mAI on the Edge
31mCounting Groceries with Computer Vision
44mArtificial Intelligence in Small Embedded Systems
26mSolving Clientelling for Luxury Goods
15mSimplifying & Scaling Edge AI