Make your app smarter
summary
On-device AI uses Huawei’s HiAI stack: HiAI Foundation compiles models (TensorFlow/Caffe2/PaddlePaddle/CoreML) into an offline NPU format supporting 140+ operators and fuses layers for bandwidth savings; Kirin 980 runs some image models 25x faster with 50x less power. HiAI Engine exposes 30+ Java APIs (face, segmentation, OCR, video summarization) plus an Android Studio plugin (DevEco) and remote testing on over 3,000 smartphones with auto-tests and performance reports.
AI-written summary · about
chapters (17)
- 0:00Introduction to smarter apps
- 2:15Challenges of AI in apps
- 2:58Introducing HiAI platform
- 4:03On-device AI development flow
- 6:09Hardware acceleration with NPU
- 8:46NPU performance benchmarks
- 9:32Real-time segmentation demo
- 10:12Third-party app success stories
- 11:59HiAI Engine pre-built APIs
- 14:10Using AI APIs in Kotlin
- 15:56WPS Office vision use case
- 17:00StorySign accessibility app
- 18:27HiAI IDE plugin overview
- 21:16Remote device testing labs
- 23:28Future SDK developments
- 25:16Market adoption and alternatives
- 27:46Resources and contact info
related talks
24mMake your app smarter
37mEnabling AI For Everyone - Hiai Open Platform Introduction
01h 20mWorkshop1: AI for Everyone
14mAccelerating On-Device AI Development
23m2025: On‑Device AI Goes Mainstream on Android
43mWhat’s new in Android machine learning
54mHarnessing On-Device AI to Transform the Smartphone Experience
10mAI in your pocket: Building intelligent Android apps