Cloud Machine Learning APIs for fun and profit with Harry Potter
summary
A working demo uses a Raspberry Pi door (camera, mic, speaker, servo with a paper clip lock) that sends audio and images to Google Cloud: Speech API for streaming and batch STT (custom phrase hints, Mandarin demo), Vision API for labels, landmark geolocation, OCR, logo and face detection (detection only, not recognition), Natural Language API for entities/sentiment, Translate and Video Intelligence (annotates shots, ~4s video processed/sec). Free tiers: Natural Language 5,000 units/month (1 unit = 1,000 Unicode chars); Vision ~1,000 requests/month per feature.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 0:23Harry Potter
- 3:28Home Security
- 3:55Vision Speech
- 4:41Systems
- 5:29Answering questions
- 6:06What is machine learning
- 7:12Why machine learning has taken off
- 7:28GPUs
- 8:34Networking
- 9:16Cloud as a supercomputer
- 10:23Google Drive
- 11:10Google Cloud
- 12:17Speech API
- 17:27Vision API
- 18:53Landmark Detection
- 20:03Face Detection
- 31:14Sentences
- 35:23Free tiers
- 42:52Crazy combinations
- 46:22Demo
related talks
59mMachine Learning and Big Data Magic with Node.js and Google Cloud
42mMachine Learning APIs, from Training [...]
57mBuilding smarter solutions with Machine Learning, from magic to reality
47mBuilding smarter apps with Machine Learning, from magic to reality
36mClosing Keynote: Machine Learning APIs with PHP
36mCloud AI Services: What they are and how to use them
59mMachine Intelligence at Google Scale
01h 00mBoost your App with Machine Learning APIs