# 树莓派 YOLO 检测

**Repository Path**: hardychenlong/raspberry-pie4b-yolo-detection

## Basic Information

- **Project Name**: 树莓派 YOLO 检测
- **Description**: 树莓派 YOLO 循环检测检测
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 15
- **Forks**: 4
- **Created**: 2020-11-27
- **Last Updated**: 2025-11-13

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README





# 项目概述 Abstract


在树莓派4B等**嵌入式设备**进行 [Yolo-Fastest](https://github.com/dog-qiuqiu/Yolo-Fastest) **推理部署**

[gitee](https://gitee.com/hardychenlong/raspberry-pie4b-yolo-detection)

# 系统需求 environment

环境：树莓派4B 内存4G版 存储tf卡32GB，系统 Ubuntu MATE 20LTS，opencv4.4.0，python3.8


# 安装说明 Installation

## 树莓派4B环境 OpenCV 安装

[Ubuntu 树莓派安装 OpenCV4  参考blog 请注意时效性！！！](https://blog.csdn.net/qq_44880154/article/details/108849953)


```bash
git clone https://gitee.com/hardychenlong/raspberry-pie4b-yolo-detection.git
```

# 使用说明 Usage

```py
#  单张图片测试
python main_yolov3.py --image=person.jpg
```



```py
# url = "http://192.168.137.136:8085/?action=stream"
url = "http://192.168.137.136:8080/?action=stream"
# 打开网络摄像头
cap = cv.VideoCapture(url)
# cap = cv.VideoCapture(0) # 电脑默认摄像头
# cap = cv.VideoCapture('dance.mp4')
```
- 修改 camera.py 程序中的摄像头读取方式，可以是摄像头或者视频或者是网络视频流。

![image-20201127212511713](https://gitee.com/hardychenlong/giteeblogimg/raw/master/20201127212518.png)
![](Asava_pic/object.jpg)

# 数据集 Dataset



如果想使用自己的数据集训练模型，可以参考作者仓库中关于 train 的部分，使用darknet 框架训练

- https://github.com/dog-qiuqiu/Yolo-Fastest#how-to-train

- https://github.com/dog-qiuqiu/Yolo-FastestV2#how-to-train


# 模型下载 Model Download

由于模型体积较小，本仓库直接包含了的模型可以直接使用，也可以选择到作者仓库下载最新的模型进行替换。

- https://github.com/dog-qiuqiu/Yolo-Fastest

# 示例

`main_yolov3.py` 

```py
# Initialize the parameters
confThreshold = 0.25  # Confidence threshold
nmsThreshold = 0.4  # Non-maximum suppression threshold
inpWidth = 320  # Width of network's input image
inpHeight = 320  # Height of network's input image
```

- 设置输入图片大小 320 * 320
- 置信度0.25
- 极大值抑制指标0.4



```py
# Load names of classes
classesFile = "voc.names"
```
- 模型使用 [VOC2007](http://host.robots.ox.ac.uk/pascal/VOC/) / coco2017 数据集预训练，识别类别参考 `voc.names` / `coco.names`
- 可以选择加载 voc 或 coco 的标签 labels
- 注意需要模型和 labels 保持一致，如 选择了voc的模型，就需要加载 voc.names 


```py
# Give the configuration and weight files for the model and load the network using them.
modelConfiguration = "Yolo-Fastest-voc/yolo-fastest-xl.cfg"
modelWeights = "Yolo-Fastest-voc/yolo-fastest-xl.weights"
```

- 配置网络结构和加载权重，可以选择 voc 或 coco 的预训练网络和权重，识别类型数量分别是20、80 , 网络结构 cfg 文件可查看 `Yolo-Fastest-coco/yolo-fastest-xl.cfg` / `Yolo-Fastest-voc/yolo-fastest-xl.cfg`





```py
net = cv.dnn.readNetFromDarknet(modelConfiguration, modelWeights)
net.setPreferableBackend(cv.dnn.DNN_BACKEND_OPENCV)
net.setPreferableTarget(cv.dnn.DNN_TARGET_CPU)
```
- 使用OpenCV dnn模块加载 darknet 网络，并使用CPU进行推理
> OpenCV的dnn模块是一个深度神经网络模块，可用于对图像、视频进行分类、识别、检测等任务。它支持各种常见的神经网络框架，如Caffe、TensorFlow、Torch等，并提供了一组简单易用的API，方便用户在OpenCV中使用深度学习模型。


#  Issue template

```md
## Description

[Insert a brief description of the issue here.]

## Steps to Reproduce

1. [Step 1]
2. [Step 2]
3. [Step 3, etc.]

## Expected Behavior

[Describe what you expected to happen.]

## Actual Behavior

[Describe what actually happened.]

## Screenshots (if applicable)

[Add any relevant screenshots here.]

## Additional Context

[Add any additional information that might be helpful here.]
```

# 许可证信息 license 

MIT License

Copyright (c) [2020] [hardychenlong]

# 参考文献/资料 Refs

- Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 779-788.
- Redmon J, Farhadi A. YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 7263-7271.
- Redmon J, Farhadi A. Yolov3: An incremental improvement[J]. arXiv preprint arXiv:1804.02767, 2018.
- https://github.com/dog-qiuqiu/Yolo-Fastest
- https://blog.csdn.net/nihate/article/details/108670542
- https://blog.csdn.net/qq_44880154/article/details/128554830?spm=1001.2014.3001.5501