yolo5生成engine jetson nano
时间: 2023-08-11 22:05:59 浏览: 88
生成Yolov5的engine文件在Jetson Nano上的步骤如下:
1. 将生成的.wts文件复制到Jetson Nano上的yolov5文件夹中。这个文件夹的路径是yolov5-5.0(Tensorrtx)\tensorrtx-yolov5-v5.0\yolov5。\[1\]
2. 打开yololayer.h文件,并根据你训练模型的类别数量修改CLASS_NUM的值。例如,如果你的模型有55个类别,就将CLASS_NUM设置为55。\[1\]
3. 在yolov5文件夹中打开终端,并依次运行以下指令:
```
mkdir build
cd build
cmake ..
make
sudo ./yolov5 -s ../yolov5s.wts yolov5s.engine
```
4. 生成的engine文件将会在yolov5文件夹中。\[1\]
接下来,将生成的yolov5s.engine和libmyplugin.so文件拷贝到DeepStream的目录中。具体步骤如下:
1. 进入DeepStream的源代码目录:
```
cd /opt/nvidia/deepstream/deepstream-5.1/sources/
```
2. 复制yolov5s.engine和libmyplugin.so文件到DeepStream的Yolov5-in-Deepstream-5.0/Deepstream_5.0/目录中:
```
cp /home/nano/tensorrtx/yolov5/build/yolov5s.engine /opt/nvidia/deepstream/deepstream-5.1/sources/Yolov5-in-Deepstream-5.0/Deepstream_5.0/
cp /home/nano/tensorrtx/yolov5/build/libmyplugin.so /opt/nvidia/deepstream/deepstream-5.1/sources/Yolov5-in-Deepstream-5.0/Deepstream_5.0/
```
3. 编译DeepStream:
```
sudo chmod -R 777 /opt/nvidia/deepstream/deepstream-5.1/sources/
git clone https://github.com/Glory-Peng/Yolov5-in-Deepstream-5.0.git
```
完成上述步骤后,你将在DeepStream的Yolov5-in-Deepstream-5.0/Deepstream_5.0/目录中得到生成的yolov5s.engine和libmyplugin.so文件,这两个文件在后续的使用中非常重要。\[2\]\[3\]
#### 引用[.reference_title]
- *1* [Jetson Nano部署YOLOv5与Tensorrtx加速——(自己走一遍全过程记录)](https://blog.csdn.net/Mr_LanGX/article/details/128094428)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^koosearch_v1,239^v3^insert_chatgpt"}} ] [.reference_item]
- *2* *3* [yolov5在jetson nano上的部署 deepstream](https://blog.csdn.net/Pcl2001/article/details/125957727)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^koosearch_v1,239^v3^insert_chatgpt"}} ] [.reference_item]
[ .reference_list ]