# age

**Repository Path**: sunplus-group/age

## Basic Information

- **Project Name**: age
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 2
- **Created**: 2025-02-07
- **Last Updated**: 2025-04-25

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

### 介绍
C3V NPU code，提供python接口，非交叉编译，请在C3V平台上编译

### 安装，编译，测试

#### 1.安装必要的软件包
```
sudo apt install swig ccache libopencv-dev libeigen3-dev libboost-all-dev python3-dev
```

#### 2.编译
```
make
```


#### 3.测试
##### a.通用测试
```
python test_vnn.py <model> <input> <flag>
```
- model : 加载模型文件，提供可用的在 \<project dir\>/models 目录下
- input : image or video
- flag  : 2 代表 print nn result (shape and numpy data)  

如果遇见类似错误
> Traceback (most recent call last):  
File "test_vnn.py", line 3, in <module>  
import vnn  
File "/home/sunplus/age/vnn.py", line 12, in <module>  
import _vnn  
ImportError: /home/sunplus/miniconda/envs/lanenet/bin/../lib/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by /home/sunplus/age/_vnn.so)  

请执行
```
conda install -c conda-forge libstdcxx-ng
```
##### b.LaneNet 测试
基于reference: https://github.com/MaybeShewill-CV/lanenet-lane-detection , 优化 DBscan

fish shell下配置PYTHONPATH(其他shell自行修改)，为了能找到编译生成的 vnn 库
```
set -x PYTHONPATH=$PYTHONPATH:.
python project/lanenet/test_lanenet.py models/lanenet_16.nb project/lanenet/data/tusimple_test_image/0.jpg 2
```