qq44642754a / yolov8_ros Goto Github PK

If I want to use my own weights, can I run the official YOLOv8 training script on Windows and copy the generated weight file into your code for use

if i want use cuda

  <param name="use_cpu"           value="true" />

change value false? or cuda?

我最开始时直接pip安装的pytorch官网的torch版本，这个是可以实现目标检测的，但是帧率只有1帧，我发现这是因为我torch.cuda.is_available()为false，所以使用了cpu来计算。后来将pytorch版本改为了nvidia官方提供的pytorch版本，这样运行帧率提高了，因为挂载到了cuda上，但是显示的图像并没有运行目标检测，这是因为什么呢

How do I use my training set?

I am facing the issue when launching the yolo_v8 launch file. It shows cannot import all the msg file. Any idea? Please help

Traceback (most recent call last):
File "/home/user/catkin_ws/devel/lib/yolov8_ros/yolo_ros.py", line 15, in
exec(compile(fh.read(), python_script, 'exec'), context)
File "/home/user/catkin_ws/src/yolov8_ros/yolov8_ros/scripts/yolo_ros.py", line 14, in
from yolov8_ros_msgs.msg import BoundingBox, BoundingBoxes, DepthPoint, DepthPoints , Object, ObjectLocations
ImportError: cannot import name 'BoundingBox' from 'yolov8_ros_msgs.msg' (unknown location)
[yolov8_node-1] process has died [pid 64501, exit code 1, cmd /home/user/catkin_ws/devel/lib/yolov8_ros/yolo_ros.py __name:=yolov8_node __log:=/home/user/.ros/log/883ecf8a-85d2-11ee-b80d-1f706cd9acc6/yolov8_node-1.log].
log file: /home/user/.ros/log/883ecf8a-85d2-11ee-b80d-1f706cd9acc6/yolov8_node-1*.log
all processes on machine have died, roslaunch will exit
shutting down processing monitor...
... shutting down processing monitor complete
done

yolov8s.pt运行闪退，就换yolov8n.pt；在yolo_v8.py强制使用cuda