Human Detector for ROS 2

This ROS 2 package utilizes MediaPipe and depth images to detect the position of a human in the x, y, and z coordinates.

The package has been tested with the RealSense D435i camera along with the corresponding Gazebo classic plugin.

Results

Simulation

Visible point cloud is added only for visualization purposes and it is not part of package.

Real world example

Setup

Clone the Repository

mkdir -p ~/ros2_ws/src
cd ~/ros2_ws/src
git clone https://github.com/Wiktor-99/human_detector.git
cd ..
rosdep install --from-paths src --ignore-src -r -y

Build

colcon build --symlink-install --cmake-args -DCMAKE_BUILD_TYPE=Release

Running

To start, open a terminal and execute the following command:

ros2 run human_detector human_detector

Next, open a second terminal and enter the following commands:

ros2 lifecycle set /human_detector configure
ros2 lifecycle set /human_detector activate

Interface

Input

The node utilizes the following topics as inputs:

• /camera/depth/image_rect_raw (Image) - for capturing the RGB image
• /camera/depth/image_rect_raw (Image) - for capturing the depth image
• /camera/depth/camera_info (CameraInfo) - for capturing camera information

Constraints

There is one strong limitation rgb image and depth image must be same size.

Output

The node publishes the following data:

• Transform to detected human
• image_with_detected_human (Image) - for publishing the image with the detected human if publish_image_with_detected is set to true.

Parameters

The node uses the following parameters:

• camera_frame_id - Name of the frame containing the camera. Default is set to camera_link.
• detected_human_frame_id - Name of the frame containing the detected human. Default is set to detected_human.
• detected_human_transform_frequency - Frequency of publishing the transform to the detected human. Default is set to 10Hz.
• publish_image_with_detected - If Topic with detected human should be published or not. If true image with detected person will be published on the image_with_detected_human topic.
• min_detection_confidence - Mediapipe parameter. Minimum confidence value ([0.0, 1.0]) from the person-detection model for the detection to be considered successful. Default to 0.5
• min_tracking_confidence - Mediapipe parameter. Minimum confidence value ([0.0, 1.0]) from the landmark-tracking model for the pose landmarks to be considered tracked successfully, or otherwise person detection will be invoked automatically on the next input image.

Devcontainer

For development and running package devcontainer can be used. All dependencies already installed via in the container.