Errors when testing tools

[AleaScovill]

<style> </style>

List of tools	Object detection and tracking of humans in the field	Scene understanding	Animal detection	Test output
rosrun perception fall_detection.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x			Output: [INFO] [1663159449.682438, 167824.240000]: Fall detection node started!
rosrun perception pose_estimation.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x			Output: [INFO] [1663159294.174351, 167687.850000]: Pose estimation node started!
rosrun perception object_tracking_2d_fair_mot.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x			Got this continuous output: error in modulated_deformable_im2col_cuda: too many resources requested for launch error in modulated_deformable_im2col_cuda: too many resources requested for launch [INFO] [1663159159.210091, 167569.950000]: Published annotated image [INFO] [1663159159.214582, 167569.950000]: Published detection boxes [INFO] [1663159159.218589, 167569.950000]: Published tracking ids
rosrun perception object_tracking_2d_deep_sort.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x			Can’t find this part
rosrun perception object_detection_2d_ssd.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x	x	x	Detection: Yes/No - Not all scenario is detected. Speed: 1.28 FPS (output) @ 4.76 FPS (input) - Moving average size = 400
rosrun perception object_detection_2d_yolov3.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x	x	x	Detection: Yes Speed: 1.44 FPS (output) @ 4.76 FPS (input) - Moving average size = 400
rosrun perception object_detection_2d_centernet.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x	x	x	Detection: Yes - Multiple Bbox for 1 object. Speed: 1.31 FPS (output) @ 4.76 FPS (input) - Moving average size = 400
rosrun perception object_detection_2d_detr.py _input_image_topic:=/vision/camera/cropeye3/image_raw	x	x	x	Didn’t output anything on the topic. Last output on cmd: [INFO] [1663158288.858492, 0.000000]: DETR node started!
rosrun perception semantic_segmentation_bisenet.py _input_image_topic:=/vision/camera/cropeye3/image_raw		x		[FATAL] [1663158442.962755]: cannot register subscription topic [/_input_image_topic:=/vision/camera/cropeye3/image_raw] with master: ERROR: parameter [topic] contains illegal chars [ERROR] [1663158442.969023]: Unable to immediately register with master node [http://192.168.20.100:11311]: master may not be running yet. Will keep trying.

[tsampazk]

Hey @AleaScovill! I have some questions and comments:

1. Are you testing the updated nodes from the develop branch or are you using the master branch?
2. From what i can tell, pose estimation, fall detection and object detection 2d yolov3 are running without any problems?
3. The basic way to modify the nodes would be through argparse, so to run a node with a custom input and output topic it would look like rosrun perception pose_estimation.py -i my_input_topic -o my_output_topic, and you can also refer to rosrun perception pose_estimation.py -h for more information
4. Even if you use the most recent nodes from the develop branch, some nodes might not be updated yet, please refer to the tracker issue [Tracker] ROS1 nodes updates and fixes #305

[AleaScovill]

Hi Kostas, 1. We are testing the docker - Not sure if it is develop or master. TX2_install is the branch. Docker version tx_2_v1.1 2. Yes 3. Thanks 4. Will the documentation be updated to reflect issue #305? Best regards / Med venlig hilsen Alea Scovill R&D Manager AGROINTELLI | Agro Food Park 13, Skejby | 8200 Aarhus N | Denmark | www.agrointelli.com<http://www.agrointelli.com/> | Mobile +45 6049 7299 From: Kostas Tsampazis ***@***.***> Sent: 14. september 2022 16:11 To: opendr-eu/opendr ***@***.***> Cc: Alea Scovill ***@***.***>; Mention ***@***.***> Subject: Re: [opendr-eu/opendr] Errors when testing tools (Issue #309) Hey @AleaScovill<https://github.com/AleaScovill>! I have some questions and comments: 1. Are you testing the updated nodes from the develop branch or are you using the master branch? 2. From what i can tell, pose estimation, fall detection and object detection 2d yolov3 are running without any problems? 3. The basic way to modify the nodes would be through argparse, so to run a node with a custom input and output topic it would look like rosrun perception pose_estimation.py -i my_input_topic -o my_output_topic, and you can also refer to rosrun perception pose_estimation.py -h for more information 4. Even if you use the most recent nodes from the develop branch, some nodes might not be updated yet, please refer to the tracker issue #305<#305> — Reply to this email directly, view it on GitHub<#309 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AZ2CTCNK3V7NY6WWT2XUE6LV6HMGXANCNFSM6AAAAAAQMMUKFY>. You are receiving this because you were mentioned.

[tsampazk]

Hey @AleaScovill, thanks for answering the questions!

1. We are testing the docker - Not sure if it is develop or master. TX2_install is the branch. Docker version tx_2_v1.1

The docker image you are using is built from the master branch so it doesn't contain the latest updates of the nodes unfortunately.
We had a discussion with Pavlos regarding this today. We will investigate into building an image from the develop branch so you can test the most up-to-date nodes and other solutions which we can discuss in the next meeting.

4. Will the documentation be updated to reflect issue [Tracker] ROS1 nodes updates and fixes #305?

I'm currently in the process of overhauling the ROS README file to reflect the changes so far.