Enable IoT with Edge Computing and Machine Learning

on YouTube ↗49m

" summary

Shows how to run real ML on constrained devices by packaging models as containers, deploying via Azure IoT Edge (edgeAgent/edgeHub, module twins, manifest -> container registry -> device) and routing data between modules; demonstrates using custom vision compact model exported as TensorFlow/Docker to classify frames (cat vs dog) and transform video to small JSON events for local alerts, cutting cloud egress and latency. Notes costs (Tesla V100 $19,000, M60 $4,587, AMD $4,399), Raspberry Pi support limited to Raspbian Stretch, and that large stores still use server racks for full accuracy while edge runs reduced‑fidelity but working models.

AI-written summary · about

" chapters (6)

1. 0:00Start
2. 6:44CAAS (Cognition as a Service)
3. 20:54Azure IoT Edge Runtime
4. 31:42Routing - INTO
5. 38:21Supported Systems
6. 48:02Learning Resources

" related talks

• 31mBuild Your Intelligent EdgeGerhard Ebinger · WeAreDevelopers AI Congress Vienna 2018
• 45mFrom Data to AI on The EdgeHenk Boelman · Build Stuff 2020
• 01h 01mML and the IoT: Living on the EdgeBrandon D Satrom · NDC London 2020
• 20mHow ML is Powering the Future of IoTGayathri Jegan Mohan · Conf42 Machine Learning 2025
• 32mExtending machine learning to industrial IoT applications at the edge with AWSIan Perez Ponce · Arm AIoT Dev Summit 2019
• 01h 02mAI on the EdgeNigel Parker, Dan Glover · NDC Sydney 2018
• 59mML and the IoT: Living on the EdgeBrandon D Satrom · NDC Oslo 2019
• 55mUsing TensorFlow for Performing Deep Learning at the EdgeAditya Bhattacharya · Developers Summit 2021