Skip to content

README_EN.md

Latest commit

 

History

History
141 lines (103 loc) · 4.16 KB

README_EN.md

File metadata and controls

141 lines (103 loc) · 4.16 KB
Raw

简体中文 | English

axcl-samples

Introduction

AXCL-Samples is developed by AXERA. This project provides sample code for running common open-source deep learning algorithms on PCIE accelerator cards powered by AXERA SoCs, enabling community developers to quickly evaluate and adapt.

Supported OS

  • Ubuntu
  • Debian
  • Windows 11

Supported Boards

Board Image Chip Vendor Link
AI Core AX-M1 AX650N Radxa Docs
M4Chat AX8850 Sipeed Wiki
LLM-8850 Card AX8850 M5Stack Docs

AXCL

AXCL is a C/Python API library for developing deep neural network inference, transcoding and other applications on AXERA chip platforms. It provides APIs for runtime resource management, memory management, model loading and execution, and media data processing.

Quick Start

Build from Source

  • Ensure AXCL deb packages are properly installed following the AXCL Documentation. Header files and libraries should be installed at /usr/include/axcl/ and /usr/lib/axcl/ respectively.
  • The following example is demonstrated on Raspberry Pi 5.

Clone the Repository

git clone https://github.com/AXERA-TECH/axcl-samples.git

Install Build Tools

Install the required build tools via apt install:

sudo apt update
sudo apt install build-essential cmake libopencv-dev

Build

mkdir build && cd build
cmake ..
make install -j4

After building, the sample binaries will be generated under ./install/bin:

axera@raspberrypi:~/temp/axcl-samples/build $ tree install
install
└── bin
    ├── ax_classification
    ├── ax_depth_anything
    ├── ax_yolo11
    ├── ax_yolo11_pose
    ├── ax_yolo11_seg
    ├── ax_yolov10
    ├── ax_yolov10_u
    ├── ax_yolov5_face
    ├── ax_yolov5s
    ├── ax_yolov5s_seg
    ├── ax_yolov8
    ├── ax_yolov8_pose
    ├── ax_yolov8_seg
    ├── ax_yolov9
    └── ax_yolov9_u

Run Examples

axera@raspberrypi:~/temp/axcl-samples/build $ ./install/bin/ax_yolo11 -m yolo11x.axmodel -i ssd_horse.jpg
--------------------------------------
model file : yolo11x.axmodel
image file : ssd_horse.jpg
img_h, img_w : 640 640
--------------------------------------

input size: 1
    name:   images [unknown] [unknown]
        1 x 640 x 640 x 3


output size: 3
    name: /model.23/Concat_output_0
        1 x 80 x 80 x 144

    name: /model.23/Concat_1_output_0
        1 x 40 x 40 x 144

    name: /model.23/Concat_2_output_0
        1 x 20 x 20 x 144

==================================================

Engine push input is done.
--------------------------------------
post process cost time:1.09 ms
--------------------------------------
Repeat 1 times, avg time 43.09 ms, max_time 43.09 ms, min_time 43.09 ms
--------------------------------------
detection num: 6
17:  96%, [ 216,   71,  423,  370], horse
16:  93%, [ 144,  203,  196,  345], dog
 0:  89%, [ 273,   14,  349,  231], person
 2:  88%, [   1,  105,  132,  197], car
 0:  82%, [ 431,  124,  451,  178], person
19:  46%, [ 171,  137,  202,  169], cow
--------------------------------------

Resources

Model Resources

NPU Toolchain

NPU toolchain documentation and downloads:

  • Pulsar2 (Support AX650A/AX650N/AX630C/AX620Q)

Community

  • Github issues
  • QQ Group: 139953715