爱芯元智AX650N部署yolov5s 自定义模型

wuboy · 发表于2023-11-16 19:43

爱芯元智AX650N部署yolov5s 自定义模型 [复制链接]

<div class='showpostmsg'>> 本博客将向你展示零基础一步步的部署好自己的yolov5s模型（博主展示的是安全帽模型），利用yolov5 官方的代码工具导出onnx模型，并通过onnxsim自带的工具精简网络结构，导出子图，为了Pulsar2 工具进行处理模型做准备。

## 获得自定义训练得到的yolov5s onnx模型

### 准备自定义数据集（博主用的是VOC数据集）

- `数据集目录结构`如下：
```
└─VOC2028:        自定义数据集
  ├─Annotations    存放的是数据集标签文件，xml格式
  ├─ImageSets        数据集的划分文件
  │└─Main
  ├─JPEGImages    存放的是数据集图片
```

- `分割数据集`

在split_train_val.py文件路径下执行`python3 split_train_val.py`会得到一下目录结构：
```
└─VOC2028:        自定义数据集
  ├─Annotations    存放的是数据集标签文件，xml格式
  ├─ImageSets        数据集的划分文件
  │└─Main test.txt
    └─test.txt
    └─train.txt
    └─val.txt
  ├─JPEGImages    存放的是数据集图片
  ├─split_train_val.py    分割数据集的py文件

```

`split_train_val.py文件代码如下`：

```
# -*- coding: utf-8 -*-
"""
Author:dragonforward
简介：分训练集、验证集和测试集，按照 8：1：1 的比例来分，训练集8，验证集1，测试集1
"""
import os
import random
import argparse

parser = argparse.ArgumentParser()
# xml文件的地址，根据自己的数据进行修改 xml一般存放在Annotations下
parser.add_argument('--xml_path', default='Annotations/', type=str, help='input xml label path')
# 数据集的划分，地址选择自己数据下的ImageSets/Main
parser.add_argument('--txt_path', default='ImageSets/Main/', type=str, help='output txt label path')
opt = parser.parse_args()

train_percent = 0.8# 训练集所占比例
val_percent = 0.1  # 验证集所占比例
test_persent = 0.1 # 测试集所占比例

xmlfilepath = opt.xml_path
txtsavepath = opt.txt_path
total_xml = os.listdir(xmlfilepath)

if not os.path.exists(txtsavepath):
  os.makedirs(txtsavepath)

num = len(total_xml)
list = list(range(num))

t_train = int(num * train_percent)
t_val = int(num * val_percent)

train = random.sample(list, t_train)
num1 = len(train)
for i in range(num1):
  list.remove(train[i])

val_test = [i for i in list if not i in train]
val = random.sample(val_test, t_val)
num2 = len(val)
for i in range(num2):
  list.remove(val[i])

file_train = open(txtsavepath + '/train.txt', 'w')
file_val = open(txtsavepath + '/val.txt', 'w')
file_test = open(txtsavepath + '/test.txt', 'w')

for i in train:
  name = total_xml[i][:-4] + '\n'
  file_train.write(name)

for i in val:
  name = total_xml[i][:-4] + '\n'
  file_val.write(name)

for i in list:
  name = total_xml[i][:-4] + '\n'
  file_test.write(name)
  
  
file_train.close()
file_val.close()
file_test.close()
```

- `voc转label得到label文件`

目录结构如下：
```
└─VOC2028:        自定义数据集
  ├─Annotations    存放的是数据集标签文件，xml格式
  ├─ImageSets        数据集的划分文件
  │└─Main
  ├─JPEGImages    存放的是数据集图片
  └─labels        yolov5将此文件夹当作训练的标注文件夹
└─voc_label.py
```

`voc_label.py文件代码如下`：

```
# -*- coding: utf-8 -*-
import xml.etree.ElementTree as ET
import os

sets = ['train', 'val', 'test']# 如果你的Main文件夹没有test.txt，就删掉'test'
classes = ["hat", "people"] # 改成自己的类别，VOC数据集有以下20类别
# classes = ["brickwork", "coil","rebar"] # 改成自己的类别，VOC数据集有以下20类别
# classes = ["aeroplane", 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
#    'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor']# class names
# abs_path = os.getcwd() /root/yolov5/data/voc_label.py 
abs_path = '/root/yolov5/data/'

def convert(size, box):
  dw = 1. / (size[0])
  dh = 1. / (size[1])
  x = (box[0] + box[1]) / 2.0 - 1
  y = (box[2] + box[3]) / 2.0 - 1
  w = box[1] - box[0]
  h = box[3] - box[2]
  x = x * dw
  w = w * dw
  y = y * dh
  h = h * dh
  return x, y, w, h

def convert_annotation(image_id):
  in_file = open(abs_path + '/VOC2028/Annotations/%s.xml' % (image_id), encoding='UTF-8')
  out_file = open(abs_path + '/VOC2028/labels/%s.txt' % (image_id), 'w')
  tree = ET.parse(in_file)
  root = tree.getroot()
  size = root.find('size')
  w = int(size.find('width').text)
  h = int(size.find('height').text)
  for obj in root.iter('object'):
  difficult = obj.find('difficult').text
  # difficult = obj.find('Difficult').text
  cls = obj.find('name').text
  if cls not in classes or int(difficult) == 1:
    continue
  cls_id = classes.index(cls)
  xmlbox = obj.find('bndbox')
  b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
     float(xmlbox.find('ymax').text))
  b1, b2, b3, b4 = b
  # 标注越界修正
  if b2 > w:
    b2 = w
  if b4 > h:
    b4 = h
  b = (b1, b2, b3, b4)
  bb = convert((w, h), b)
  out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')

for image_set in sets:
  if not os.path.exists(abs_path + '/VOC2028/labels/'):
  os.makedirs(abs_path + '/VOC2028/labels/')

image_ids = open(abs_path + '/VOC2028/ImageSets/Main/%s.txt' % (image_set)).read().strip().split()
  list_file = open(abs_path + '/VOC2028/%s.txt' % (image_set), 'w')
  for image_id in image_ids:
  list_file.write(abs_path + '/VOC2028/JPEGImages/%s.jpg\n' % (image_id))# 要么自己补全路径，只写一半可能会报错
  convert_annotation(image_id)
  list_file.close()

```

![picture 1](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182450101-520375331.png)

### 训练模型

- 配置环境
```
git clone https://github.com/ultralytics/yolov5
cd yolov5
pip install -r requirements.txt
pip install onnx
```

- 下载预训练权重（博主尝试了v7.0的和v6.0的pt都可以）
```
https://github.com/ultralytics/yolov5/releases/download/v7.0/yolov5s.pt
```
![picture 0](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182450592-865793378.png)

- 训练（博主使用的是学校的集群进行训练）

```
python3 train.py --weights weights/yolov5s.pt --cfg models/yolov5s.yaml --data data/safthat.yaml --epochs 150 --batch-size 16 --multi-scale --device 0 
```
![picture 2](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182451149-439909649.png)

```
python3 detect.py --source /root/yolov5/data/images/000000.jpg --weights /root/yolov5/runs/train/exp13/weights/best.pt --conf-thres 0.25
```
![picture 3](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182451784-1715265443.png)

- 导出onnx模型，并通过onnxsim自带的工具精简网络结构，导出子图

```
python3 export.py --weightsyolov5s_hat.pt --include onnx
pip install onnx-simplifier
python3 -m onnxsim yolov5s_hat.onnx yolov5s_hat_sub.onnx
```

![picture 4](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182452270-1336271791.png)

模型查看通过网址输入netron.app查看三个输出：

![picture 8](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182452720-1788864856.png)

查看官方给的yolov5s.json文件：

```
{
"model_type": "ONNX",
"npu_mode": "NPU1",
"quant": {
  "input_configs": [
  {
  "tensor_name": "images",
  "calibration_dataset": "./dataset/calibration_dataset.tar.gz",
  "calibration_size": 4,
  "calibration_mean": [0, 0, 0],
  "calibration_std": [255.0, 255.0, 255.0]
  }
  ],
  "calibration_method": "MinMax",
  "precision_analysis": false
},
"input_processors": [
  {
  "tensor_name": "images",
  "tensor_format": "BGR",
  "src_format": "BGR",
  "src_dtype": "U8",
  "src_layout": "NHWC"
  }
],
"output_processors": [
  {
  "tensor_name": "326",
  "dst_perm": [0, 2, 3, 1]
  },
  {
  "tensor_name": "370",
  "dst_perm": [0, 2, 3, 1]
  },
  {
  "tensor_name": "414",
  "dst_perm": [0, 2, 3, 1]
  }
],
"compiler": {
  "check": 0
}
}

```

output_processors给的326，370，414

根据[秋水大佬博客](https://zhuanlan.zhihu.com/p/569083585)的导出子图文件修改得到适应pulsar2工具的onnx文件：

```
import onnx

input_path = "yolov5s_hat-sim.onnx"
output_path = "yolov5s_hat_sub.onnx"
input_names = ["images"]
output_names = ["326","370","414"]
onnx.utils.extract_model(input_path, output_path, input_names, output_names)
```

问题记录：
（1）导出子图很关键
   由于之前对深度学习知识了解很少，才知道子图，如果执行使用yolov5s_hat_sub.onnx，在Pulsar2工具中进行量化的话，会生成axmodel但是是无法执行了，通过`o0圏圏蟲0o`大佬给的模型对比网络结构和秋水大佬的ax620a部署yolov5博客才明白自己问题出现在哪里

如果你没处理导出子图得到的图会是这样：
![picture 5](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182453144-516336166.png)

正确量化后的yolov5s图：

官方的yolov5s模型：

![picture 6](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182453433-737702152.png)
博主的正确的axmodel：
![picture 7](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182453693-1387288305.png)

其中数字不同255和21的原因（秋水大佬博客有讲）：
该数字为（模型训练的类别+5）*3，博主类别为2类，所以就是21

## 模型转化（关键）

### 安装docker环境（已安装过的可以跳过）
```
安装docker依赖的基础软件
sudo apt-get update
sudo apt-get install apt-transport-https ca-certificates curl gnupg-agent software-properties-common
添加官方来源
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
安装 docker
sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io
```

### pulsar2工具模型转化以及仿真运行

- 下载pulsar2工具（博主使用的是1.9）

[谷歌网盘链接：](https://drive.google.com/file/d/1-NW7ExBXj5-nTha40iwYshjNJb74Zfer/view?usp=drive_link)

[quick_start_example 文件夹](https://github.com/AXERA-TECH/pulsar2-docs/releases/download/v1.9/quick_start_example.zip)
其中文件如下：
```
configdatasetmodeloutputpulsar2-run-helper
```

./dataset/calibration_data.tar：添加了数据集的四张照片
config文件：
```
{
"model_type": "ONNX",
"npu_mode": "NPU1",
"quant": {
  "input_configs": [
  {
  "tensor_name": "images",
  "calibration_dataset": "./dataset/calibration_data.tar",
  "calibration_size": 4,
  "calibration_mean": [0, 0, 0],
  "calibration_std": [255.0, 255.0, 255.0]
  }
  ],
  "calibration_method": "MinMax",
  "precision_analysis": false
},
"input_processors": [
  {
  "tensor_name": "images",
  "tensor_format": "BGR",
  "src_format": "BGR",
  "src_dtype": "U8",
  "src_layout": "NHWC"
  }
],
"output_processors": [
  {
  "tensor_name": "326",
  "dst_perm": [0, 2, 3, 1]
  },
  {
  "tensor_name": "370",
  "dst_perm": [0, 2, 3, 1]
  },
  {
  "tensor_name": "414",
  "dst_perm": [0, 2, 3, 1]
  }
],
"compiler": {
  "check": 0
}
}
```

```
（博主是直接就是root用户，如果不是root用户记得加sudo）
root@LAPTOP-U638FQQS:~# docker load -i ax_pulsar2_1.9_enc.tar.gz
root@LAPTOP-U638FQQS:~# docker images -a
REPOSITORY      TAG   IMAGE ID   CREATED   SIZE
hhb4tools/rv_debian_build 1.3   08f478d17c34 7 weeks ago 2.16GB
pulsar2       1.9_enc 641ba18a8da3 2 months ago  3.46GB
hhb4tools/hhb     2.4.5 58df969ae05a 3 months ago  8.16GB
hhb4tools/rv_ub20_build   1.4   a65456ded4f0 6 months ago  5.11GB
hhb4tools/march-user-static 1.0   d9efab34da5e 7 months ago  301MB
riscv64/ubuntu      22.04 8b55084b9c02 11 months ago 61.6MB
riscv64/ubuntu      21.04 d0b60ed75c22 21 months ago 60.3MB

root@LAPTOP-U638FQQS:~#docker run -it --net host --rm -v $PWD:/data pulsar2:1.9_enc

root@1657ec5355e2:/data# pulsar2 version
version: 1.9
commit: c62d0b64
root@1657ec5355e2:/data#
```
1. 编译执行
以 yolov5s_hat_sub.onnx 为例, 执行如下 pulsar2 build 命令编译生成 compiled.axmodel:
```
pulsar2 build --input model/yolov5s_hat_sub.onnx --output_dir output --config config/yolov5s_config.json
```
博主输出信息：
```
root@1657ec5355e2:/data# pulsar2 build --input model/yolov5s_hat_sub.onnx --output_dir output --config config/yolov5s_config.json
2023-11-16 17:49:21.238 | WARNING| yamain.command.build:fill_default:320 - ignore images csc config because of src_format is AutoColorSpace or src_format and tensor_format are the same
Building onnx ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00
2023-11-16 17:49:22.486 | INFO | yamain.command.build:build:444 - save optimized onnx to [output/frontend/optimized.onnx]
2023-11-16 17:49:22.489 | INFO | yamain.common.util:extract_archive:21 - extract [dataset/calibration_data.tar] to [output/quant/dataset/images]...
                 Quant Config Table                  
┏━━━━━━━━┳━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Input┃ Shape    ┃ Dataset Directory ┃ Data Format ┃ Tensor Format ┃ Mean    ┃ Std       ┃
┡━━━━━━━━╇━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━┩
│ images │ [1, 3, 640, 640] │ images    │ Image   │ BGR   │ [0.0, 0.0, 0.0] │ [255.0, 255.0, 255.0] │
└────────┴──────────────────┴───────────────────┴─────────────┴───────────────┴─────────────────┴───────────────────────┘
Transformer optimize level: 0
4 File(s) Loaded.
[17:49:24] AX LSTM Operation Format Pass Running ...  Finished.
[17:49:24] AX Set MixPrecision Pass Running ...   Finished.
[17:49:24] AX Refine Operation Config Pass Running ...  Finished.
[17:49:24] AX Reset Mul Config Pass Running ...   Finished.
[17:49:24] AX Tanh Operation Format Pass Running ...  Finished.
[17:49:24] AX Confused Op Refine Pass Running ...   Finished.
[17:49:24] AX Quantization Fusion Pass Running ...  Finished.
[17:49:24] AX Quantization Simplify Pass Running ...  Finished.
[17:49:24] AX Parameter Quantization Pass Running ... Finished.
Calibration Progress(Phase 1): 100%|███████████████████████████████████████████████████████████████████████████████| 4/4 [00:01<00:00,2.16it/s]
Finished.
[17:49:26] AX Passive Parameter Quantization Running ...Finished.
[17:49:26] AX Parameter Baking Pass Running ...   Finished.
[17:49:26] AX Refine Int Parameter Pass Running ...   Finished.
[17:49:26] AX Refine Weight Parameter Pass Running ...  Finished.
--------- Network Snapshot ---------
Num of Op:      [142]
Num of Quantized Op:    [142]
Num of Variable:    [267]
Num of Quantized Var:   [267]
------- Quantization Snapshot ------
Num of Quant Config:    [430]
BAKED:        [60]
OVERLAPPED:       [168]
ACTIVATED:      [138]
SOI:        [4]
PASSIVE_BAKED:      [60]
Network Quantization Finished.
quant.axmodel export success: output/quant/quant_axmodel.onnx
===>export input/output data to folder: output/quant/debug/test_data_set_0
Building native ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00
2023-11-16 17:49:28.704 | WARNING| yamain.command.load_model:pre_process:454 - preprocess tensor [images]
2023-11-16 17:49:28.704 | INFO | yamain.command.load_model:pre_process:456 - tensor: images, (1, 640, 640, 3), U8
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:pre_process:456 - op: op:pre_dequant_1, AxDequantizeLinear, {'const_inputs': {'x_zeropoint': array(0, dtype=int32), 'x_scale': array(1., dtype=float32)}, 'output_dtype': <class 'numpy.float32'>, 'quant_method': 0}
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:pre_process:456 - tensor: tensor:pre_norm_1, (1, 640, 640, 3), FP32
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:pre_process:456 - op: op:pre_norm_1, AxNormalize, {'dim': 3, 'mean': [0.0, 0.0, 0.0], 'std': [255.0, 255.0, 255.0]}
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:pre_process:456 - tensor: tensor:pre_transpose_1, (1, 640, 640, 3), FP32
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:pre_process:456 - op: op:pre_transpose_1, AxTranspose, {'perm': [0, 3, 1, 2]}
2023-11-16 17:49:28.705 | WARNING| yamain.command.load_model:post_process:475 - postprocess tensor [326]
2023-11-16 17:49:28.705 | INFO | yamain.command.load_model:handle_postprocess:473 - op: op:post_transpose_1, AxTranspose
2023-11-16 17:49:28.705 | WARNING| yamain.command.load_model:post_process:475 - postprocess tensor [370]
2023-11-16 17:49:28.706 | INFO | yamain.command.load_model:handle_postprocess:473 - op: op:post_transpose_2, AxTranspose
2023-11-16 17:49:28.706 | WARNING| yamain.command.load_model:post_process:475 - postprocess tensor [414]
2023-11-16 17:49:28.706 | INFO | yamain.command.load_model:handle_postprocess:473 - op: op:post_transpose_3, AxTranspose
tiling op... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 241/241 0:00:00
new_ddr_tensor = []
<frozen backend.ax650npu.oprimpl.normalize>:186: RuntimeWarning: divide by zero encountered in divide
<frozen backend.ax650npu.oprimpl.normalize>:187: RuntimeWarning: invalid value encountered in divide
build op... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1177/1177 0:00:04
add ddr swap... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1141/1141 0:00:00
calc input dependencies... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1437/1437 0:00:00
calc output dependencies... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1437/1437 0:00:00
assign eu heuristic ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1437/1437 0:00:00
assign eu onepass ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1437/1437 0:00:00
assign eu greedy ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1437/1437 0:00:00
2023-11-16 17:49:34.720 | INFO | yasched.test_onepass:results2model:2004 - max_cycle = 4,846,471
2023-11-16 17:49:35.349 | INFO | yamain.command.build:compile_npu_subgraph:1076 - QuantAxModel macs: 7,881,318,400
2023-11-16 17:49:35.352 | INFO | yamain.command.build:compile_npu_subgraph:1084 - use random data as gt input: images, uint8, (1, 640, 640, 3)
2023-11-16 17:49:38.687 | INFO | yamain.command.build:compile_ptq_model:1003 - fuse 1 subgraph(s)
root@1657ec5355e2:/data# 
```

2. 模型仿真运行

```
cp output/compiled.axmodel pulsar2-run-helper/models/yolov5s_hat.axmodel
输入数据准备
python3 cli_detection.py --pre_processing --image_path sim_images/000032.jpg --axmodel_path models/yolov5s_hat.axmodel --intermediate_path sim_inputs/0
输出信息：
root@1657ec5355e2:/data/pulsar2-run-helper# python3 cli_detection.py --pre_processing --image_path sim_images/000032.jpg --axmodel_path models/yolov5s_hat.axmodel --intermediate_path sim_inputs/0
[I] Write [images] to 'sim_inputs/0/images.bin' successfully.
仿真模型推理
pulsar2 run --model models/yolov5s_hat.axmodel --input_dir sim_inputs --output_dir sim_outputs --list list.txt
输出信息：
root@1657ec5355e2:/data/pulsar2-run-helper# pulsar2 run --model models/yolov5s_hat.axmodel --input_dir sim_inputs --output_dir sim_outputs --list list.txt
Building native ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:00
>>> [0] start
write [326] to [sim_outputs/0/326.bin] successfully
write [370] to [sim_outputs/0/370.bin] successfully
write [414] to [sim_outputs/0/414.bin] successfully
>>> [0] finish
输出数据处理（记得指定图片路径）
python3 cli_detection.py --post_processing --image_path sim_images/000032.jpg --axmodel_path models/yolov5s_hat.axmodel --intermediate_path sim_outputs/0
输出信息：
root@1657ec5355e2:/data/pulsar2-run-helper# python3 cli_detection.py --post_processing --image_path sim_images/000032.jpg --axmodel_path models/yolov5s_hat.axmodel --intermediate_path sim_outputs/0
[I] Number of detected objects: 7
[I]0: 94.59%, [972, 224, 1089, 345]
[I]0: 94.56%, [886, 222, 970, 327]
[I]0: 94.12%, [1141, 145, 1262, 281]
[I]0: 93.96%, [487, 249, 571, 370]
[I]0: 93.94%, [331, 201, 430, 341]
[I]0: 93.27%, [186, 192, 302, 351]
[I]0: 62.17%, [607, 233, 692, 338]
root@1657ec5355e2:/data/pulsar2-run-helper#
```
具体工具文档如下：
[pulsar2工具文档](https://pulsar2-docs.readthedocs.io/zh-cn/latest/user_guides_quick)

## 开发板运行

开发板镜像为1.27版本，采用本地编译

下载源码：
```
git clone https://github.com/AXERA-TECH/ax-samples.git
```
修改ax_yolov5s_steps.cc文件中：

```
修改classname标签
const char* CLASS_NAMES[] = {
  "person", "hat"};
generate_proposals_yolov5函数指定classnum数量为2
for (uint32_t i = 0; i < io_info->nOutputSize; ++i)
  {
  auto& output = io_data->pOutputs[i];
  auto& info = io_info->pOutputs[i];
  auto ptr = (float*)output.pVirAddr;
  int32_t stride = (1 << i) * 8;
  detection::generate_proposals_yolov5(stride, ptr, PROB_THRESHOLD, proposals, input_w, input_h, ANCHORS, prob_threshold_u_sigmoid,2);
  }
```
修改的原因是它默认为80，不然会报错下面信息：
```
root@maixbox:/home/ax-samples/build/install/ax650# ./ax_yolov5s -m yolov5s_hat.axmodel -i 000032.jpg
--------------------------------------
model file : yolov5s_hat.axmodel
image file : 000032.jpg
img_h, img_w : 640 640
--------------------------------------
WARN,Func(__is_valid_file),NOT find file = '/etc/ax_syslog.conf'
ERROR,Func(__syslog_parma_cfg_get), NOT find = '/etc/ax_syslog.conf'
Engine creating handle is done.
Engine creating context is done.
Engine get io info is done.
Engine alloc io is done.
Engine push input is done.
--------------------------------------
Segmentation fault

```
![picture 9](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182454221-175286927.png)

```

cd ax-samples 
mkdir build && cd build
cmake -DBSP_MSP_DIR=/soc/ -DAXERA_TARGET_CHIP=ax650 ..
make -j6
make install
```
编译完成后，生成的可执行示例存放在 `ax-samples/build/install/ax650/` 路径下：

```bash
ax-samples/build$ tree install
install
└── ax650
  ├── ax_classification
  ├── ax_detr
  ├── ax_dinov2
  ├── ax_glpdepth
  ├── ax_hrnet
  ├── ax_imgproc
  ├── ax_pfld
  ├── ax_pp_humanseg
  ├── ax_pp_liteseg_stdc2_cityscapes
  ├── ax_pp_ocr_rec
  ├── ax_pp_person_attribute
  ├── ax_pp_vehicle_attribute
  ├── ax_ppyoloe
  ├── ax_ppyoloe_obj365
  ├── ax_realesrgan
  ├── ax_rtmdet
  ├── ax_scrfd
  ├── ax_segformer
  ├── ax_simcc_pose
  ├── ax_yolo_nas
  ├── ax_yolov5_face
  ├── ax_yolov5s
  ├── ax_yolov5s_seg
  ├── ax_yolov6
  ├── ax_yolov7
  ├── ax_yolov7_tiny_face
  ├── ax_yolov8
  ├── ax_yolov8_pose
  └── ax_yolox
```
讲axmodel模型放在可执行文件下和测试图片：
```
root@maixbox:/home/ax-samples/build/install/ax650# ./ax_yolov5s -m yolov5s_hat.axmodel -i 000032.jpg
--------------------------------------
model file : yolov5s_hat.axmodel
image file : 000032.jpg
img_h, img_w : 640 640
--------------------------------------
WARN,Func(__is_valid_file),NOT find file = '/etc/ax_syslog.conf'
ERROR,Func(__syslog_parma_cfg_get), NOT find = '/etc/ax_syslog.conf'
Engine creating handle is done.
Engine creating context is done.
Engine get io info is done.
Engine alloc io is done.
Engine push input is done.
--------------------------------------
post process cost time:0.42 ms
--------------------------------------
Repeat 1 times, avg time 6.15 ms, max_time 6.15 ms, min_time 6.15 ms
--------------------------------------
detection num: 7
 0:95%, [ 981,221, 1080,342], person
 0:95%, [ 332,201,431,341], person
 0:95%, [ 886,222,970,327], person
 0:94%, [1141,140, 1262,290], person
 0:94%, [ 187,197,300,347], person
 0:94%, [ 487,252,571,373], person
 0:91%, [ 605,232,689,337], person
--------------------------------------
```
![picture 10](https://img2023.cnblogs.com/blog/2915785/202311/2915785-20231116182455655-1136582421.png)

## 感谢
感谢`o0圏圏蟲0o`,`无事闲来`，`梦醒时分`，`N/A`，`秋水`等大佬的帮助</div><script>					var loginstr = '<div class="locked">查看本帖全部内容，请<a href="javascript:;"   style="color:#e60000" class="loginf">登录</a>或者<a href="https://bbs.eeworld.com.cn/member.php?mod=register_eeworld.php&action=wechat" style="color:#e60000" target="_blank">注册</a></div>';
					
					if(parseInt(discuz_uid)==0){
						
					}		</script><script type="text/javascript">(function(d,c){var a=d.createElement("script"),m=d.getElementsByTagName("script")[0],eewurl="//counter.eeworld.com.cn/pv/count/";a.src=eewurl+c;m.parentNode.insertBefore(a,m)})(document,523)</script>

爱芯元智AX650N部署yolov5s 自定义模型 [复制链接]

浏览过的版块