# Estimate human poses with `sample_hrnet_pose_estimation`
This pose estimation sample application uses the `sample_hrnet_pose_estimation` ROS node to provide high-precision human pose estimation capabilities.
It processes input images and publishes the following ROS 2 topics:
- `/pose_estimation_results`: Output images with visualized pose keypoints.
- `/pose_estimation_points`: Raw keypoint coordinates in a structured message format.
Note
- For model information, see [HRNetPose - Qualcomm AI Hub](https://aihub.qualcomm.com/iot/models/hrnet_pose?searchTerm=hrnet).
- For the latest sample information, see the [sample_hrnet_pose_estimation](https://github.com/qualcomm-qrb-ros/qrb_ros_samples/tree/main/ai_vision/sample_hrnet_pose_estimation) GitHub repository.

## Pipeline flow for `sample_hrnet_pose_estimation`
## ROS nodes used in `sample_hrnet_pose_estimation`
| ROS node | Description |
| --- | --- |
| [qrb_ros_camera](https://github.com/qualcomm-qrb-ros/qrb_ros_camera) | The QRB ROS Camera is a ROS package that publishes the images from Qualcomm CSI and GMSL cameras. |
| [qrb_ros_nn_interface](https://github.com/qualcomm-qrb-ros/qrb_ros_nn_inference) | `QRB_ROS_NN_inference` is a ROS 2 package that performs neural network model, providing AI-based perception for robotics applications. |
| [image_publisher_node](https://github.com/ros-perception/image_pipeline) | The image\_publisher is a ROS 2 Jazzy package that publishes the image ROS topic with local path. |
| `hrnet_pose_estimation_node` | The node subscribes to `/image_raw` for image input, performs pose estimation, and publishes the results to `/pose_estimation_results` and `/pose_estimation_points` topics. |
## ROS topics used in `sample_hrnet_pose_estimation`
| ROS topic | Type | Description |
| --- | --- | --- |
| `/image_raw` | `sensor_msgs.msg.Image` | Indicates the input image to `sample_hrnet_pose_estimation`. |
| `/qrb_inference_input_tensor` | `qrb_ros_tensor_list_msgs/msg/TensorList` | The preprocessed image is converted into an input message for nn inference. |
| `/qrb_inference_output_tensor` | `qrb_ros_tensor_list_msgs/msg/TensorList` | Indicates the message after the nn inference. |
| `/pose_estimation_results` | `sensor_msgs.msg.Image` | Indicates the output images with visualized pose keypoints. |
| `/pose_estimation_points` | `geometry_msgs.msg.PolygonStamped` | Indicates the raw keypoint coordinates in a structured message format. |
Note
`/pose_estimation_points` contains the coordinates of 17 key points on the original image. Developers can subscribe to this topic for secondary development. The key points include: nose, left eye, and right eye. The specific correspondence is as follows:
| ID | Body part | ID | Body part |
| --- | --- | --- | --- |
| 1 | Nose | 10 | Right wrist |
| 2 | Right eye | 11 | Left wrist |
| 3 | Left eye | 12 | Right hip |
| 4 | Right ear | 13 | Left hip |
| 5 | Left ear | 14 | Right knee |
| 6 | Right shoulder | 15 | Left knee |
| 7 | Left shoulder | 16 | Right ankle |
| 8 | Right elbow | 17 | Left ankle |
| 9 | Left elbow | | |
## Prerequisites
You have completed the settings in [Set up the environment for running sample applications](https://docs.qualcomm.com/doc/80-70033-265/topic/quick_start.html#setup-demo-qs).
## Run out-of-the-box `sample_hrnet_pose_estimation`
1. On the development kit, run the following commands:
# Set HOME variable
(ssh) export HOME=/opt
# Remount the /usr directory with read-write permissions
(ssh) mount -o remount rw /usr
# Set up the runtime environment
(ssh) source /usr/share/qirp-setup.sh -m
# Set the ROS_DOMAIN_ID
(ssh) export ROS_DOMAIN_ID=123
# Launch the sample with an image publisher, You can replace 'image_path' with the path to your desired image.
(ssh) ros2 launch sample_hrnet_pose_estimation launch_with_image_publisher.py image_path:=/usr/share/sample_hrnet_pose_estimation/input_image.jpg
# Launch the sample hrnet pose estimation node with qrb_ros_camera ros node.
(ssh) ros2 launch sample_hrnet_pose_estimation launch_with_qrb_ros_camera.py
Copy to clipboard
2. In the Docker environment of the host computer, run the following commands:
1. Start the `rqt`.
# YOUR_HOST_IP is the IP address of the Host where you want to view the sample output.
(ssh) export DISPLAY=YOUR_HOST_IP:0
(ssh) export ROS_DOMAIN_ID=123
(ssh) rqt
Copy to clipboard
2. Select the following buttons in sequence:
Plugins --> Visualization --> Image View
3. Select `pose_estimation_results` to see the results.
4. Open a new terminal and run the following command to view the `/pose_estimation_points` topic.
source /usr/share/qirp-setup.sh -m
export ROS_DOMAIN_ID=123
ros2 topic echo /pose_estimation_points
Copy to clipboard
## Build and run `sample_hrnet_pose_estimation`
1. In a terminal of the host computer, run the following commands:
1. Build the sample application project.
cd /target/qcs9075-iq-9075-evk/qirpsdk_artifacts/qcs9075-iq-9075-evk
tar -zxvf qirp-sdk_2.6.2.tar.gz && cd qirp-sdk
source setup.sh
# build sample
cd /qirp-samples/ai_vision/sample_hrnet_pose_estimation
colcon build
Copy to clipboard
2. Package and push the sample application to the device.
# package and push build result of sample
cd /qirp-samples/ai_vision/sample_hrnet_pose_estimation/install/sample_hrnet_pose_estimation
tar -czvf sample_hrnet_pose_estimation.tar.gz lib share
scp sample_hrnet_pose_estimation.tar.gz root@[ip-addr]:/opt/
Copy to clipboard
2. On the development kit, run the following commands:
1. Install the sample application.
# Remount the /usr directory with read-write permissions
(ssh) mount -o remount rw /usr
# Install sample package
(ssh) tar --no-overwrite-dir --no-same-owner -zxf /opt/sample_hrnet_pose_estimation.tar.gz -C /usr/
Copy to clipboard
2. Run the sample application with the steps in [Run out-of-the-box sample\_hrnet\_pose\_estimation](https://docs.qualcomm.com/doc/80-70033-265/topic/sample_hrnet_pose_estimation.html#run-out-of-the-box-hrnet-pose-estimation).
Last Published: Sep 30, 2026
[Previous Topic
Detect faces with sample\_face\_detection](https://docs.qualcomm.com/bundle/publicresource/80-70033-265/topics/sample_face_detection.md) [Next Topic
Detect objects with sample\_object\_detection](https://docs.qualcomm.com/bundle/publicresource/80-70033-265/topics/sample_object_detection.md)