Skip to content

Latest commit

 

History

History
62 lines (51 loc) · 3.07 KB

RunWithROS.md

File metadata and controls

62 lines (51 loc) · 3.07 KB
Raw

Object Detection, Segmentation, and Tracking Using a Camera Sensor Image Topic

This document provides a guide on how to run the ROS_IST given the image topic published as sensor_msgs/Image.msg by a camera sensor. First, make sure that the Anaconda environment where the ROS_IST is installed is initialized prior to running the scripts below.

Next, check that ROS master is running or run the following command to start the ROS master in a terminal:

roscore

Object Detector Node

Next, create a new terminal and run the following command to start the ROS node that will subscribe to the camera image topic:

python src/segment_and_track_ros.py --input_topic /image_topic

Please specify the ROS topic of the image in --input_topic. Additionally, you can use --camera_param to input a resized or cropped image to the model to speed up the process.

A blank Tkinter window for GUI will pop up.

Next, to initialize the GUI for target selection, initialize the rosservice in another terminal:

rosservice call /init_segmentation "{}"

The target selection GUI will appear. Please select the targets by following this workflow. For detailed instructions on using the GUI, please refer to GUI.md.

  1. Select one target at a time using the following operations:
    • Draw a bounding box around the object by clicking and dragging the mouse.
    • Right-click on the object to specify it as foreground or background. You can change the mode using button.
  2. Click the "Generate Masks" button to display and register the masked image.
    • If you are not satisfied with the results, use the "Reselect" or "Clear All" button to remove them.
  3. Repeat steps 1-2 for each object you want to track.
  4. After selecting all targets, click the "Start Tracking" button to initiate the tracking process.

Once tracking starts, masks and their positions will be published via /masks and /tracked_targets topics. You can check the output by checking in terminal, i.e.:

rostopic echo /masks

rostopic echo /tracked_targets

Visualizing

We provide two ways to visualize the results: masked images and images with points indicating the pixel positions (2D) of the tracked targets.

Run the following command in a new terminal to start the ROS node that subscribes to the results from segment_and_tracking_ros.py and visualizes them:

python src/visualizer.py --input_topic /image_topic

You can visualize the images in two ways by specifying command-line arguments:

  1. Masked image (default)
  • This is the default visualization mode. To disable it, add the --disable_mask_mode argument when running visualizer.py.
  1. Image with points
  • You can visualize the positions of the tracked targets by adding the --point_mode argument when running visualizer.py. This mode displays points at the tracked target positions.

For example, to run the visualizer with point mode enabled in addition to mask mode, use the following command:

python src/visualizer.py --point_mode

To view the visualized image, run the following command in another terminal:

rosrun rqt_image_view rqt_image_view