Skip to content

Prometheus目标检测算法 目标检测 点击跟踪

Ren Jin edited this page Mar 31, 2022 · 2 revisions

1 使用教程

Prometheus目标检测+点击跟踪使用教程如下图所示:

tutorial_tracking

具体请参考相应链接:

2 编译与安装

注意,运行Yolov5-TensorRT需要安装CUDA、TensorRT,如果没有安装,请参考附录1

先看速度测试:

平台 Yolov5模型 输入分辨率 检测帧率
Nvidia GTX 1080Ti yolov5s 608 65.2
Nvidia Jetson Xavier NX yolov5s 608 27.1

测试视频-bilibili

1. 下载编译prometheus-yolov5-tensorrt子模块

首先安装2个依赖项:

# 安装Opencv
sudo apt-get install libopencv-dev
# 安装pycuda
pip3 install pycuda

在Prometheus目录下,输入:

cd <path-to-prometheus>
chmod +x ./Scripts/install_detection_yolov5tensorrt.sh
./Scripts/install_detection_yolov5tensorrt.sh

或者输入以下命令:

git clone https://gitee.com/jario-jin/prometheus-yolov5-tensorrt.git Modules/object_detection_yolov5tensorrt
cd Modules/object_detection_yolov5tensorrt
pip3 install -r requirements.txt

下载训练好的权重文件,以官方权重yolov5s.pt为例,用如下代码生成yolov5s.wts文件

cd <path-to-prometheus>/Modules/object_detection_yolov5tensorrt
python3 gen_wts.py

编译tensorrtx-yolov5:

cd <path-to-prometheus>/Modules/object_detection_yolov5tensorrt
mkdir build
cd build
cmake ..
make
sudo ./yolov5 -s # 生成yolov5s.engine文件

2. 运行yolov5-tensorrt

启动yolov5-tensorrt服务端(负责读取摄像头,检测目标)

cd <path-to-prometheus>
chmod +x ./Scripts/start_yolov5tensorrt_server.sh
./Scripts/start_yolov5tensorrt_server.sh

启动yolov5-tensorrt客户端(负责从服务端读取检测结果,并发布ros消息)

roslaunch prometheus_detection yolov5_nvidia_tensorrt.launch

注意:

  • 点击鼠标左键开始跟踪目标,点击鼠标右键取消跟踪,恢复到检测模式

附录1

1. 安装CUDA

以CUDA10.0为例,进入CUDA下载官网https://developer.nvidia.com/cuda-10.0-download-archive,选择Linux→x86_64→Ubuntu→18.04→deb(local)下载.deb格式安装包,然后运行以下指令:

cd Downloads
sudo dpkg -i cuda-repo-ubuntu1804-10-0-local-10.0.130-410.48_1.0-1_amd64.deb
sudo apt-key add /var/cuda-repo-10-0-local-10.0.130-410.48/7fa2af80.pub
sudo apt-get update
sudo apt-get install cuda

如果遇到依赖错误,可采用aptitude安装:

sudo apt-get install aptitude
sudo aptitude install cuda
2. 安装TensorRT

以Tensorrt7.0.0为例,进入Tensorrt下载官网https://developer.nvidia.com/nvidia-tensorrt-7x-download,选择Tensorrt 7.0.0.11 for Ubuntu 1804 and CUDA 10.0 DEB local repo packages下载.deb格式安装包,然后运行以下指令:

sudo dpkg -i nv-tensorrt-repo-ubuntu1804-cuda10.0-trt7.0.0.11-ga-20191216_1-1_amd64.deb
sudo apt-get update
sudo apt-get install tensorrt
Clone this wiki locally