Make your app smarter
summary
Tells whether on-device AI actually runs: HiAI explains running models on an NPU with 25x speedup and 50x lower power, shows model conversion and one‑bit quantization, and publishes encrypted models; demonstrates a model store, an Android Studio plugin and a dev board; claims real‑time rates (about 4,000 images per minute), APIs for face detection, OCR, segmentation and super‑resolution, and free remote device testing four hours per day with automated compatibility and battery/performance reports.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:24HOW SMART YOUR PHONE IS
- 3:07A BETTER USER EXPERIENCE
- 4:09REAL TIME AI SEGMENTATION
- 4:51A DEDICATED SET OF AL INSTRUCTIONS
- 6:02CHALLENGES
- 6:40BUILD AN ECOSYSTEM
- 7:37HIAI APPLICATION DEVELOPMENT FLOW
- 11:13HIAI ENGINE APIS
- 12:26IMAGE SEMANTIC SEGMENTATION
- 12:59DOCUMENT RECOGNITION
- 13:10FACE RECOGNITION
- 14:27INTEGRATE A CUSTOM MODEL
- 16:27OPEN LABS
- 16:46COMPATIBILITY TEST
- 17:24PERFORMANCE TEST
- 20:55NEXT SUCCESSFUL CASE
- 21:46WHAT'S NEXT
- 23:43JOIN THE ALLIANCE
related talks
29mMake your app smarter
14mAccelerating On-Device AI Development
01h 20mWorkshop1: AI for Everyone
47mMyths and Reality of Android Machine Learning
37mEnabling AI For Everyone - Hiai Open Platform Introduction
31mBringing AI on-Device: From Cloud to Edge
23m2025: On‑Device AI Goes Mainstream on Android
43mWhat’s new in Android machine learning