一、安装依赖包

sudo apt install \
libssl1.0.0 \
libgstreamer1.0-0 \
gstreamer1.0-tools \
gstreamer1.0-plugins-good \
gstreamer1.0-plugins-bad \
gstreamer1.0-plugins-ugly \
gstreamer1.0-libav \
libgstrtspserver-1.0-0 \
libjansson4=2.11-1

二、下载并安装deepstream sdk

1）下载DeepStream 5.1 Jetson tar package deepstream_sdk_v5.1.0_jetson.tbz2
#链接：https://pan.baidu.com/s/1LGP1QMwWf0ZTauXAO4pu4Q    提取码：80ij 
2）解压并安装
sudo tar -xvf deepstream_sdk_v5.1.0_jetson.tbz2 -C /
cd /opt/nvidia/deepstream/deepstream-5.1
sudo ./install.sh
sudo ldconfig

三、在电脑端完成模型pt转wts

（此处务必检查yolov5与tensorrtx包的版本）
git clone -b v5.0 https://github.com/ultralytics/yolov5.git
git clone -b yolov5-v5.0 https://github.com/wang-xinyu/tensorrtx.git
cp {tensorrtx}/yolov5/gen_wts.py {ultralytics}/yolov5
cd {ultralytics}/yolov5
python gen_wts.py -w best.pt -o best.wts

四、在jetson上完成wts转engine

1）在jetson上克隆v5.0版本tensorrtx
git clone -b yolov5-v5.0 https://github.com/wang-xinyu/tensorrtx.git
2）编译
#检查yolov5.cpp中的模型权重NET，默认的是s
#检查yololayer.h中CLASS_NUM的数量，默认的是80
cd tensorrtx/yolov5
mkdir build
cd build
cmake ..
make
3）复制best.wts到tensorrtx/yolov5/build目录下
4）生成yolov5s.engine
sudo ./yolov5 -s best.wts yolov5s.engine s
sudo ./yolov5 -d yolov5s.engine ../samples

五、DeepStream部署yolov5模型

1）设置权限
sudo chmod -R 777 /opt/nvidia/deepstream/deepstream-5.1/sources/
2）拷贝项目文件yolov5.zip到/opt/nvidia/deepstream/deepstream-5.1/sources/并解压
#链接：https://pan.baidu.com/s/1DX_DtoJMbwugGVEaWuF1bw    提取码：q0e2 
#该文件对应yolov5-v5.0，若需使用其他版本，自行修改其中yololayer.h、yololayer.cu等文件
3）编译
#修改config_infer_primary.txt中的num
#修改labels.txt
##如要改变NMS_THRESH, 编辑文件nvdsinfer_custom_impl_Yolo/nvdsparsebbox_Yolo.cpp
##如要改变CONF_THRESH, 编辑文件config_infer_primary.txt
cd /opt/nvidia/deepstream/deepstream-5.1/sources/yolov5
CUDA_VER=10.2 make -C nvdsinfer_custom_impl_Yolo

六、测试

1）测试视频文件推理
cd /opt/nvidia/deepstream/deepstream-5.1/sources/yolov5
deepstream-app -c deepstream_app_config.txt
2）USB摄像头测试
deepstream-app -c source1_usb_dec_infer_yolov5.txt
3）CSI摄像头测试
deepstream-app -c source1_csi_dec_infer_yolov5.txt