harmony 鸿蒙AI Development

  • 2023-10-30
  • 浏览 (638)

AI Development

Overview

Built on the native AI framework provided by OpenHarmony, the AI subsystem consists of the following components: - MindSpore Lite: an AI inference framework that provides unified APIs for AI inference. - Neural Network Runtime (NNRt): an intermediate bridge that connects the inference framework and AI hardware.

MindSpore Lite

MindSpore Lite is a built-in AI inference framework of OpenHarmony. It provides AI model inference capabilities for different hardware devices and end-to-end AI model inference solutions for developers to empower intelligent applications in all scenarios. Currently, MindSpore Lite has been widely used in applications such as image classification, target recognition, facial recognition, and character recognition.

Figure 1 Development process for MindSpore Lite model inference mindspore workflow

The MindSpore Lite development process consists of two phases:

  • Model conversion

MindSpore Lite uses models in .ms format for inference. You can use the model conversion tool provided by MindSpore Lite to convert third-party framework models, such as TensorFlow, TensorFlow Lite, Caffe, and ONNX, into .ms models. For details, see Converting Models for Inference.

  • Model inference

You can call the MindSpore Lite runtime APIs to implement model inference. The procedure is as follows:

  1. Create an inference context by setting the inference hardware and number of threads.
  2. Load the .ms model file.
  3. Set the model input data.
  4. Perform model inference, and read the output.

MindSpore Lite is built in the OpenHarmony standard system as a system component. You can develop AI applications based on MindSpore Lite in the following ways:

Neural Network Runtime

Neural Network Runtime (NNRt) functions as a bridge to connect the upper-layer AI inference framework and bottom-layer acceleration chip, implementing cross-chip inference computing of AI models.

MindSpore Lite supports configuration of the NNRt backend, and therefore you can directly configure MindSpore Lite to use the NNRt hardware. The focus of this topic is about how to develop AI applications using MindSpore Lite. For details about how to use NNRt, see Connecting the Neural Network Runtime to an AI Inference Framework.

你可能感兴趣的鸿蒙文章

harmony 鸿蒙AI

harmony 鸿蒙Using MindSpore Lite JavaScript APIs to Develop AI Applications

harmony 鸿蒙Using MindSpore Lite Native APIs to Develop AI Applications

0  赞