In your requirement.txt you install the python-opencv 4,
But when building cv_birdge, the CMakelist require opencv3
Should I install the opencv3 again?
I tried to modify the CMakeList from opencv3 to opencv4, but it seems to be incompatible.

I have seen your previous reply to another questioner ，And I got this error after changing the common.py file when I was operating .If it is convenient for you, I don't know if you can make a tutorial for changing the weight of YOLOV5 v6.1 .I don't know how to proceed at the moment .Thanks for reading and looking forward to your recovery .

Hi, I already trained my own weights and changed the <arg name="weights" default="best.pt"/> in detector.launch using my own weights. but the error occur. if there anything that i need to change?

The error are as shown below

[INFO] [1672724360.280203]: start detect node [INFO] [1672724360.308326]: Launched node for object detection Load model: /home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/weights/best.pt Traceback (most recent call last): File "/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py", line 304, in <module> dm = detectManager() File "/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py", line 98, in __init__ self.warmup() File "/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py", line 162, in warmup self.model = attempt_load(self.weights, map_location=self.device) # load FP32 model File "/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/models/experimental.py", line 118, in attempt_load ckpt = torch.load(w, map_location=map_location) # load File "/home/aiman/anaconda3/lib/python3.9/site-packages/torch/serialization.py", line 592, in load return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args) File "/home/aiman/anaconda3/lib/python3.9/site-packages/torch/serialization.py", line 851, in _load result = unpickler.load() AttributeError: Can't get attribute 'MP' on <module 'models.common' from '/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/models/common.py'> [detect_manager-1] process has died [pid 13174, exit code 1, cmd /home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py __name:=detect_manager __log:=/home/aiman/.ros/log/424d5572-8b24-11ed-84cd-b0fc36e80577/detect_manager-1.log]. log file: /home/aiman/.ros/log/424d5572-8b24-11ed-84cd-b0fc36e80577/detect_manager-1*.log [detect_manager-1] restarting process
Thank you

我的设备是jetson agx xavier ，melodic，配置完了环境，开始运行launch文件
roslaunch ros_pytorch_yolov5 detector.launch device:=cpu view_img:=True
roslaunch ros_pytorch_yolov5 detector.launch device:=0 view_img:=True
使用cpu和gpu都试了，ros启动正常，开始detect节点后，运行完load model，执行fusing layers....
然后就没动静了，不动了，求大佬指导。

[ERROR] [1645192563.522846]: bad callback: <bound method detectManager.imageCb of <main.detectManager object at 0x7f065f7710>>
Traceback (most recent call last):
File "/workspace/pytorch_yolov5_catkin_ws/devel/lib/python3/dist-packages/ros_pytorch_yolov5/msg/_BoundingBoxes.py", line 165, in serialize
buff.write(_get_struct_d4q().pack(_x.probability, _x.xmin, _x.ymin, _x.xmax, _x.ymax))
TypeError: only integer tensors of a single element can be converted to an index

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/topics.py", line 882, in publish
self.impl.publish(data)
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/topics.py", line 1066, in publish
serialize_message(b, self.seq, message)
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/msg.py", line 152, in serialize_message
msg.serialize(b)
File "/workspace/pytorch_yolov5_catkin_ws/devel/lib/python3/dist-packages/ros_pytorch_yolov5/msg/_BoundingBoxes.py", line 167, in serialize
except TypeError as te: self._check_types(ValueError("%s: '%s' when writing '%s'" % (type(te), str(te), str(locals().get('_x', self)))))
File "/opt/ros/noetic/lib/python3/dist-packages/genpy/message.py", line 393, in _check_types
check_type(n, t, getattr(self, n))
File "/opt/ros/noetic/lib/python3/dist-packages/genpy/message.py", line 314, in check_type
check_type(field_name + '[]', base_type, v)
File "/opt/ros/noetic/lib/python3/dist-packages/genpy/message.py", line 324, in check_type
check_type('%s.%s' % (field_name, n), t, getattr(field_val, n))
File "/opt/ros/noetic/lib/python3/dist-packages/genpy/message.py", line 273, in check_type
raise SerializationError('field %s must be float type' % field_name)
genpy.message.SerializationError: field bounding_boxes[].probability must be float type

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/topics.py", line 750, in invoke_callback
cb(msg)
File "/workspace/pytorch_yolov5_catkin_ws/src/ros_pytorch_yolov5/yolov5/detect.py", line 120, in imageCb
detections = self.detect(self.cv_image, data)
File "/workspace/pytorch_yolov5_catkin_ws/src/ros_pytorch_yolov5/yolov5/detect.py", line 365, in detect
self.pub.publish(detection_results)
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/topics.py", line 886, in publish
raise ROSSerializationException(str(e))
rospy.exceptions.ROSSerializationException: field bounding_boxes[].probability must be float type

Hi!
When I run your program, the wrong color appears. Do you know the reason? The rgb information of the camera looks correct in rviz, but the color is reversed in yolo. I am using a realsense D455 camera.
Looking forward to your reply！

119行图片预处理函数的返回input_image并不是主要的推理函数detect()的输入。那这样预处理的作用是什么？同时经过我反复测试确定，imagePreProcessing()函数中的 input_img = resize(self.padded_image, (self.network_img_size, self.network_img_size, 3))/255.这一行操作极其耗时，在我的10代i7rtx2060上光是这一行函数执行都需要大约56ms，加上推理25ms，一帧的检测时间达到80-90ms。
同时存在大量多余的代码，读起来困难。并无冒犯之意。
还是感谢您的开源!

May I know why this error occur?

aiman@aiman:~/catkin_ws_cv_bridge$ roslaunch ros_pytorch_yolov5 detector.launch device:=cpu view_img:=True
... logging to /home/aiman/.ros/log/13894466-7e0f-11ed-a8b6-b0fc36e80577/roslaunch-aiman-29519.log
Checking log directory for disk usage. This may take a while.
Press Ctrl-C to interrupt
Done checking log file disk usage. Usage is <1GB.

started roslaunch server http://aiman:33769/

SUMMARY

PARAMETERS

• /detect_manager/agnostic_nms: True
• /detect_manager/augment: True
• /detect_manager/classes: None
• /detect_manager/conf_thres: 0.25
• /detect_manager/detected_objects_topic: detected_objects_...
• /detect_manager/detections_image_topic: detections_image_...
• /detect_manager/detections_results_topic: detections_result...
• /detect_manager/device: cpu
• /detect_manager/exist_ok: True
• /detect_manager/image_topic: /usb_cam/image_raw
• /detect_manager/img_size: 640
• /detect_manager/iou_thres: 0.45
• /detect_manager/name: exp
• /detect_manager/project: runs/result
• /detect_manager/publish_image: True
• /detect_manager/save_conf: True
• /detect_manager/save_txt: False
• /detect_manager/source: data/images
• /detect_manager/view_img: True
• /detect_manager/weights: yolov5s.pt
• /rosdistro: melodic
• /rosversion: 1.14.13

NODES
/
detect_manager (ros_pytorch_yolov5/detect.py)

ROS_MASTER_URI=http://localhost:11311

process[detect_manager-1]: started with pid [29537]
Traceback (most recent call last):
File "/home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py", line 14, in
from skimage.transform import resize
File "/home/aiman/anaconda3/lib/python3.9/site-packages/skimage/init.py", line 125, in
from .util.dtype import (img_as_float32,
File "/home/aiman/anaconda3/lib/python3.9/site-packages/skimage/util/init.py", line 17, in
from ._map_array import map_array
File "/home/aiman/anaconda3/lib/python3.9/site-packages/skimage/util/_map_array.py", line 2, in
from ._remap import _map_array
File "skimage/util/_remap.pyx", line 1, in init skimage.util._remap
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject
[detect_manager-1] process has died [pid 29537, exit code 1, cmd /home/aiman/catkin_ws_pytorch_yolov5/src/ros_pytorch_yolov5/yolov5/detect.py __name:=detect_manager __log:=/home/aiman/.ros/log/13894466-7e0f-11ed-a8b6-b0fc36e80577/detect_manager-1.log].
log file: /home/aiman/.ros/log/13894466-7e0f-11ed-a8b6-b0fc36e80577/detect_manager-1*.log

Hello, I used your project, but there is no picture. Can you tell me more specifically? Thank you for your help. If you can, you can communicate via email. My email is [email protected]

cd src/ros_pytorch_yolov5/bag_dataset
运行到这一步发现没有bag，尴尬了呀

i will really appreciate that , cz it rings:
from ros_pytorch_yolov5.msg import BoundingBox, BoundingBoxes
ModuleNotFoundError: No module named 'ros_pytorch_yolov5'

Hello! My English is not very good. I encountered a bad call problem when using d435i. I modified the camera resolution and topic name. Have you encountered it? How can I modify model and data? thank you.

你好，我想使用自己训练的权重来进行识别，但是我不知道都需要修改哪些地方，请问该怎么去修改呢

Hi!
I have been trying to understand the wrapper's working and can't seem to find how to choose a model type. We want to use yolov5n(nano) and don't know what the default config is. Any help regarding how to choose a model would be great. Thank you!