[Tracker] ROS1 nodes updates and fixes

[tsampazk]

This issue serves to track the progress on the updates on all ROS1 nodes, as per the use-case partner feedback as well as fixes regarding consistency across all nodes. Also, the delay issues mentioned in #275 should be fixed along with the other changes, as well as issue #209 relating to input topics etc.

Main issues to fix:

Below there's a list of main issues that need fixing, i tried to check out all the nodes and made some comments in parentheses on what i noticed, so use them to get a general sense on the issues:

1. Argparse usage as seen here.
2. Argument names and default values consistency across nodes, as seen in node class ctor and argparse. Arguments especially in argparse which is the interface the users interact with, should have descriptive names like input_rgb_image_topic and input_depth_image_topic, etc, as well as default values to run out-of-the-box if possible.
3. Updated device selection as seen here.
4. Preferably, if the node has multiple outputs, we should provide the option to disable them as seen here, with appropriate changes in the callback too.
5. Make sure you update your docstrings for wrong argument names, typos, some nodes mentioning different tools, etc.
6. Please test whether your node has delay issues as described in ROS nodes delay #275 and if you need to adjust the queue sizes.
7. Keep in mind the feedback provided by the use-case partners.
8. Use a lambda function for argparse type as seen here to handle passing of None values to disable topics. This only applies to topics that can be disabled like these ones. This gives the option to users to entirely disable some unused topic from publishing, to save resources.

Tracking list:

• Face detection retinaface

• Face recognition

• Pose estimation

• Fall detection

• Object detection 2D - centernet

• Object detection 2D - ssd

• Object detection 2D - yolov3

• Semantic segmentation bisenet

• Panoptic segmentation efficient ps

• Object detection 2D - detr (some updates in Ros2 detr #296, should be good to go. Note that changes for ROS1 are present in ros2 branch)

• Video activity recognition (argparse, device code, argument naming/default values, docstring, see rosnode - video_activity_recognition.py - parameters + error #302) (updates in Fix ros1 video_activity_recognition.py #322)

• RGBD hand gesture recognition (argparse, device code, argument naming/default values, docstrings, see rosnode - rgbd_hand_gesture_recognition.py - parameter #303) (updates in Ros1 fixed rgbd hand gestures recognition #325)

• Heart anomaly detection (updates in argparse, device code, device argument, docstrings) (updates in Ros1 fixes heart anomaly detection #330)

• Speech command recognition (updates in argparse, device code, argument naming/default values, docstring) (updates in Ros1 fixes speech command recognition #331)

• Audiovisual emotion recognition (updates in argparse, device code, device argument, argument naming/default values) (updates in Ros1 fixes audiovisual emotion recognition #332)

• Landmark-based facial expression recognition (argparse, device code, argument naming/default values, docstrings, pose_estimation variables) (updates in ros1_node for landmark_expression_recognition updated #307)

• Skeleton-based action recognition (argparse, device code, argument naming/default values, see rosnode - skeleton_based_action_recognition.py - not executable + message error #304) (updates in Har ros1 fix #308)

• Object detection 2D - gem (argparse, device code, argument naming/default values) (updates in Ros2 gem #295)

• Object detection 3D - voxel (argparse, device code, argument naming/default values, docstrings) (Ros1 fixes for 3D Detection and 2D/3D tracking #320)

• Object tracking 2D - deep sort (argparse, device code, argument naming/default values) (Ros1 fixes for 3D Detection and 2D/3D tracking #320)

• Object tracking 2D - fair mot (argparse, device code, argument naming/default values) (Ros1 fixes for 3D Detection and 2D/3D tracking #320)

• Object tracking 3D - ab3dmot (argparse, device code, argument naming/default values, docstrings) (Ros1 fixes for 3D Detection and 2D/3D tracking #320)

• End-to-end planner - (End-to-end planner updates #349)

For any questions or doubts feel free to ask below!

Edit: Added 8. in the list of issues to fix.
Edit: ROS1 GEM node updates are merged into the ros2 branch, related PR #256

[negarhdr]

Two new branches (on top of develop) for skeleton-based_action_recognition and landmark-based_facial_expression_recognition are created which are called "har_ros1_fix", "fer_ros1_fix", respectively.
In "har_ros1_fix", the script "skeleton_based_action_recognition" is updated and in "fer_ros1_fix", the script "landmark_based_facial_expression_recognition" is updated.
Please check them out and let me know if there are still problems.

[tsampazk]

Hey @negarhdr!

I checked the changes out, they seem pretty good apart from some really minor things.

Feel free to open pull requests to merge them into the develop branch and i can comment on the minor issues on the review.

[tsampazk]

I am not yet closing this issue as i am going to go through all the listed ROS1 nodes, make sure that nothing related to this issue was missed and apply any minor fixes that might be needed.