Neural Network API

NNAPI, TensorFlow Lite and GStreamer on QCS610

NNAPI, an Android C API, is designed for running machine learning operations, especially inference, on Android devices. Refer to the Machine Learning section of the NDK Guides for details on the NNAPI runtime, programming model, and programming flow.

TFLite model loading via NNAPI
The Qualcomm® QCS610 supports running TFLite models on Qualcomm® Hexagon™ DSP, Qualcomm® Adreno™ GPU and Qualcomm® Kryo™ CPU via NNAPI.
The NNAPI framework has been ported from Android to run models on our Hexagon DSP. The model is trained in TensorFlow, then frozen and converted to TFLite format. Then the model is given to the NNAPI runtime, which has the capability to offload models to the Hexagon DSP. For TFLite use cases, the GStreamer TFLite plugin can be used. The result from post-processing is handed off as machine learning metadata to the GStreamer buffer. Using NNAPI, the model can be accelerated by developing an application directly.

For more information, see the Qualcomm Neural Processing SDK for AI and Tensorflow sections of the Qualcomm® Robotics RB5 Software Reference Manual.

Qualcomm QCS610, Qualcomm Hexagon, Qualcomm Robotics RB5 and Qualcomm Neural Processing SDK are products of Qualcomm Technologies, Inc. and/or its subsidiaries.