ROS nodes FPS performance measurements #419
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Implementation wise I think this is a nice way of doing it. I have some doubts concerning whether measuring just the infer method is enough though, because in that case I don't expect there to be any real difference between the value obtained from this method and the values already reported in the documentation when the tools were benchmarked using the python API. It's measuring the same thing. So the question is, can we actually use these values to "show" that the performance requirements of the use-case have been met? Because by taking only the infer method, it sounds to me like it's an evaluation of the tool, not a task/use-case. What else should be included and where the boundary should be I don't know.. but before doing the change for all nodes it should be settled |
Oh i was under the impression that that was what we wanted (measure only tool inference). I'll wait for more input before moving forward. |
…ome rearrangement
tsampazk
requested review from
passalis,
omichel,
ad-daniel and
stefaniapedrazzi
as code owners
projects/opendr_ws/src/opendr_perception/scripts/object_detection_2d_detr_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws/src/opendr_perception/scripts/object_detection_2d_gem_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws_2/src/opendr_perception/opendr_perception/object_detection_2d_detr_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws_2/src/opendr_perception/opendr_perception/object_detection_2d_gem_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws/src/opendr_perception/scripts/object_detection_3d_voxel_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws_2/src/opendr_perception/opendr_perception/object_tracking_3d_ab3dmot_node.py
Outdated
Show resolved
Hide resolved
Co-authored-by: Stefania Pedrazzi <[email protected]>
Co-authored-by: Stefania Pedrazzi <[email protected]>
stefaniapedrazzi
previously approved these changes
Mar 27, 2023
4 tasks
passalis
reviewed
Apr 4, 2023
projects/opendr_ws/src/opendr_perception/scripts/performance_node.py
Outdated
Show resolved
Hide resolved
projects/opendr_ws/src/opendr_perception/scripts/performance_node.py
Outdated
Show resolved
Hide resolved
Co-authored-by: Nikolaos Passalis <[email protected]>
lucamarchionni
pushed a commit
to lucamarchionni/opendr
that referenced
this pull request
Jun 10, 2024
* Added performance node * Added time measurement for tool inference and publishing * Moved performance start time before image preprocessing * audiovisual_emotion_recognition_node.py performance and some formatting * binary_high_resolution_node.py performance * continual_skeleton_based_action_recognition_node.py performance and some rearrangement * face_detection_retinaface_node.py performance * face_recognition_node.py performance * facial_emotion_estimation_node.py performance * fall_detection_node.py performance * heart_anomaly_detection_node.py performance * hr_pose_estimation_node.py performance * landmark_based_facial_expression_recognition_node.py performance * object_detection_2d_centernet_node.py performance * object_detection_2d_detr_node.py performance * object_detection_2d_gem_node.py performance * object_detection_2d_nanodet_node.py performance * object_detection_2d_ssd_node.py performance * object_detection_2d_yolov3_node.py performance * object_detection_2d_yolov5_node.py performance * object_detection_3d_voxel_node.py performance * object_tracking_2d_deep_sort_node.py performance * object_tracking_2d_fair_mot_node.py performance * object_tracking_2d_siamrpn_node.py performance * object_tracking_3d_ab3dmot_node.py performance * panoptic_segmentation_efficient_lps_node.py performance * panoptic_segmentation_efficient_ps_node.py performance * pose_estimation_node.py minor fixes * rgbd_hand_gesture_recognition_node.py performance * semantic_segmentation_bisenet_node.py performance * skeleton_based_action_recognition_node.py performance * speech_command_recognition_node.py performance * video_activity_recognition_node.py performance * Added section for utility nodes and entry for the performance node * Added entry in notes for logging performance * Added entries for the new performance topic in all nodes * audiovisual_emotion_recognition_node.py ROS2 performance * binary_high_resolution_node.py ROS2 performance * continual_skeleton_based_action_recognition_node.py ROS2 performance * face_detection_retinaface_node.py ROS2 performance * face_recognition_node.py ROS2 performance * facial_emotion_estimation_node.py ROS2 performance * fall_detection_node.py ROS2 performance * heart_anomaly_detection_node.py ROS2 performance * hr_pose_estimation_node.py ros1 renamed class * hr_pose_estimation_node.py ROS2 performance * landmark_based_facial_expression_recognition_node.py ROS2 performance * object_detection_2d_centernet_node.py ROS2 performance * object_detection_2d_detr_node.py ROS2 performance * object_detection_2d_gem_node.py ROS2 performance * object_detection_2d_nanodet_node.py ROS1 minor fix * object_detection_2d_nanodet_node.py ROS2 performance * object_detection_2d_ssd_node.py ROS2 performance * object_detection_2d_yolov3_node.py ROS1 class name fix * object_detection_2d_yolov3_node.py ROS2 performance * object_detection_2d_yolov5_node.py ROS1 class name fix * object_detection_2d_yolov5_node.py ROS2 performance * object_detection_3d_voxel_node.py ROS2 performance * object_tracking_2d_deep_sort_node.py ROS2 performance * object_tracking_2d_fair_mot_node.py ROS2 performance * object_tracking_2d_siamrpn_node.py ROS2 performance * object_tracking_3d_ab3dmot_node.py ROS2 performance * panoptic_segmentation_efficient_lps_node.py ROS1 some imports rearrangement * panoptic_segmentation_efficient_lps_node.py ROS2 performance * panoptic_segmentation_efficient_ps_node.py ROS1 some imports rearrangement * panoptic_segmentation_efficient_ps_node.py ROS2 performance * pose_estimation_node.py ROS2 performance * rgbd_hand_gesture_recognition_node.py ROS2 performance * semantic_segmentation_bisenet_node.py ROS2 performance * skeleton_based_action_recognition_node.py ROS2 performance and some refactoring * speech_command_recognition_node.py ROS2 performance * video_activity_recognition_node.py ROS2 performance * Added ROS2 performance_node.py * ROS2 readme updates for performance topics/node * Apply suggestions from code review Co-authored-by: Stefania Pedrazzi <[email protected]> * Disable running tracking infer if no tracking is needed for ab3dmot ROS2 Co-authored-by: Stefania Pedrazzi <[email protected]> * Disable running tracking infer if no tracking is needed for ab3dmot ROS1 * Apply suggestions from code review Co-authored-by: Nikolaos Passalis <[email protected]> * Removed additional spaces after suggested change * Applied review suggestion to ROS2 node as well --------- Co-authored-by: Stefania Pedrazzi <[email protected]> Co-authored-by: Nikolaos Passalis <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds time performance measurement of tools' inference. For each node, the time it takes to run (only) inference is measured and published in a performance topic. Publishing this message is optional, i.e. the relevant topic needs to be set via argparse. This message can be subscribed to, or echoed to show the current FPS.
If a more accurate performance measurement is needed, one can use the new performance node which subscribes to the performance topic and calculates a running average of the FPS measurement which it prints in the console along with the time it took for the last frame to be processed.
@thomaspeyrucain @ad-daniel Let me know how this looks and whether it's sufficient and/or convenient, before i add it in the rest of the ROS1/2 nodes.