AXERA-TECH/ax-samples

Samples code for world class Artificial Intelligence SoCs for computer vision applications.

What it solves

This project provides reference implementations for running common deep learning models on AXERA-TECH's AI SoC (System-on-Chip) hardware. It bridges the gap between open-source AI algorithms and specialized edge-computing hardware, allowing developers to quickly evaluate performance and adapt models for deployment.

How it works

The repository contains C++ source code examples organized by specific chip architecture. Each example demonstrates the full pipeline: pre-processing, NPU (Neural Processing Unit) inference, post-processing, and result output. Developers use CMake-based cross-compilation to build these programs for their target hardware, utilizing the provided model conversion tools (Pulsar2) to prepare models for the NPU.

Who it’s for

This project is designed for embedded systems engineers and AI developers working with AXERA-TECH hardware who need to deploy computer vision and deep learning models on edge devices, automotive chips, or development boards like the AXera-Pi Pro.

Highlights

  • Supports a wide range of AXERA-TECH chips, including AX650, AX637, AX630C, AX620Q, AX615, and AX8860.
  • Includes ready-to-use code for common tasks like image classification, object detection, segmentation, and pose estimation.
  • Provides structured benchmark data to help developers evaluate model inference speed on specific hardware.
  • Offers comprehensive documentation for cross-compilation and integration with the Pulsar2 model toolchain.

Related

  • Project
    luxonis/oak-examples

    A collection of demonstrations and tutorials for Luxonis OAK devices, showcasing AI vision, depth measurement, and neural network implementation using DepthAI.

  • Project
    huaweicloud/ModelArts-Lab

    A collection of example codes and datasets for the Huawei Cloud ModelArts platform, enabling both no-code automated learning and full-cycle AI model development.

  • Project
    qualcomm/ai-hub-models

    Qualcomm AI Hub Models is a Python package and CLI that provides a catalog of pre‑optimised machine‑learning models (vision, language, generative AI) ready for deployment on Snapdragon CPUs, GPUs, and NPUs. It handles quantisation, compilation, and performance profiling on real Qualcomm devices via the AI Hub Workbench, and includes sample apps and demos for both local PyTorch inference and on‑device accelerated runtimes.

  • Project
    Liquid4All/cookbook

    A collection of tutorials, code examples, and deployment guides for building edge-native AI applications using Liquid Foundation Models and the LEAP SDK.

  • Project
    awslabs/awsome-distributed-ai

    A collection of reference architectures and examples for deploying and operating distributed AI training and inference on AWS services like SageMaker HyperPod and EKS.