Install XenseSDK

Installation Guide

Step 1: Prepare the Python Development Environment

# 进入 Xense SDK 目录
cd xensesdk

# 创建并激活虚拟环境
conda create -n xenseenv python=3.9
#or conda create -n xenseenv python=3.10

conda activate xenseenv

Note

It is recommended to use Anaconda with Python version 3.9 or 3.10.

Step 2: Install CUDA Toolkit and cuDNN

The SDK requires onnxruntime_gpu, along with配套的 cudnn and cudatoolkit. Select the following installation method based on your environment:

Option 1: onnxruntime_gpu>1.18.0

Install the required version:

# 这个例子使用 cuda12.9
conda install nvidia/label/cuda-12.9.0::cuda-toolkit nvidia::cudnn

Add the CUDA path to the environment variable ‘LD_LIBRARY_PATH’:

# linux 里可以运行如下命令
export LD_LIBRARY_PATH=$CONDA_PREFIX/lib:$CONDA_PREFIX/lib64 #（临时）
mkdir -p $CONDA_PREFIX/etc/conda/activate.d && echo 'export LD_LIBRARY_PATH=$CONDA_PREFIX/lib:$CONDA_PREFIX/lib64:$LD_LIBRARY_PATH' > $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh #（永久）

Option 2: onnxruntime_gpu==1.18.0 (Not available for 50 series graphics cards)

Search for the required version:

conda search cudnn
conda search cudatoolkit

Install the required version:

conda install cudnn==8.9.2.26 cudatoolkit==11.8.0

Step 3: Install the Xense SDK Package (install the SDK package into your environment):

# 如果是从本地目录安装
pip install xensesdk-0.1.0-cp39-cp39-win_amd64.whl # (对于定制软件包)
# 或者从 PyPI 安装
pip install xensesdk -i https://repo.huaweicloud.com/repository/pypi/simple/