Making Android Apps with Intelligence
summary
Shows how to add concrete capabilities: use Google Play services Mobile Vision API for face detection (not identification), barcode and OCR that run in-app; call cloud endpoints like Google Cloud Vision or Microsoft Emotion API which return JSON emotion scores per face; use TensorFlow transfer learning to retrain Inception v3 final layer with ~500–4,000 images to classify new categories and run models on-device (classify/detect/stylize) so the app works offline. Notes demo apps (mimicker alarm, Pepper+IBM Watson) and privacy, cost, and permission tradeoffs.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:59Artificial Intelligence everywhere!
- 2:19What is Machine Learning?
- 6:40Face detector
- 7:34Mobile Vision API - Barcode Detection
- 8:19REST API calls
- 10:16Cloud ML services demo
- 14:06Demo - Mimicker Alarm
- 23:02TensorFlow on mobile
- 26:41Microsoft LUIS
- 32:49Pepper the Robot + Android
- 34:50Design considerations
- 36:38What is next?
related talks
33mMaking Android Apps with Intelligence
52mThrough the Looking-Glass Eyes of an Android
31mSupercharging Android Apps with Tensorflow
30mMachine Learning in Android Apps with ML Kit
47mMyths and Reality of Android Machine Learning
40mApplied TensorFlow in Android apps
19mOn-device machine learning: TensorFlow on Android
21mLive Coding: Adding Machine Learning to Android Things in 10 Mins