Working on Android ML at Google, including TFLite for Play Services
Adrien Couque
Engineer @Google
Office Hour: Machine Learning
Come discuss Machine Learning and more with: Adrien Couque, Engineer @Google
Talk Title
How to use hardware acceleration for Machine Learning inference on Android
The rising processing power of Android devices unlocks on-device machine learning capabilities unthinkable a few years ago.
To enable machine learning real-time experiences (augmented reality filters, live speech transcription, etc…) you need to accelerate ML processes. TensorFlow Lite on Android can leverage specialized hardware available on the device (GPU, DSP, NPU, etc…).
In this session we will explain how to use Acceleration delegates and new TensorFlow Lite Android APIs to enable ML hardware acceleration.
#tflite #ML #AI #gpu #performance