Tutorial for People Segmentation with Segformer and TensorRT 
=============================================================

Overview
------------

This tutorial walks you through a graph for :ir_repo:`Image Segmentation <isaac_ros_image_segmentation>`
for segmenting people in an image using TensorRT.

Tutorial Walkthrough
--------------------

1.  Complete the :ref:`quickstart <repositories_and_packages/isaac_ros_image_segmentation/isaac_ros_segformer/index:quickstart>` up until model preparation step.
2.  Launch the Docker container using the ``run_dev.sh`` script:

    .. code:: bash

       cd ${ISAAC_ROS_WS}/src/isaac_ros_common && \
         ./scripts/run_dev.sh

3.  Run the following launch files to start the inferencing:

    .. code:: bash

        ros2 launch isaac_ros_segformer isaac_ros_people_sem_segformer_tensor_rt.launch.py engine_file_path:=${ISAAC_ROS_WS}/isaac_ros_assets/models/peoplesemsegformer/1/model.plan

4.  Open **another** terminal and play the rosbag:

    .. code:: bash

        cd ${ISAAC_ROS_WS}/src/isaac_ros_common && \
            ./scripts/run_dev.sh

    .. code:: bash

        ros2 bag play -l isaac_ros_assets/isaac_ros_segformer/segformer_sample_data --remap image_rect:=image camera_info_rect:=camera_info

5.  In another terminal, visualize and validate the output of the package by launching
    ``rqt_image_view``:

    .. code:: bash

        cd ${ISAAC_ROS_WS}/src/isaac_ros_common && \
           ./scripts/run_dev.sh

    Then launch ``rqt_image_view``:

    .. code:: bash

        ros2 run rqt_image_view rqt_image_view

    Then inside the ``rqt_image_view`` GUI, change the topic to ``/segformer/colored_segmentation_mask`` to view a colorized segmentation mask.

    .. figure:: :ir_lfs:`<resources/isaac_ros_docs/repositories_and_packages/isaac_ros_image_segmentation/isaac_ros_segformer/peoplesemsegformer_output_rqt.png>`
        :width: 320px
        :align: center

    .. note::

        The raw segmentation is also published to
        ``/segformer/raw_segmentation_mask``. However, the raw pixels correspond
        to the class labels and so the output is unsuitable for human visual
        inspection.