rzv_object_detection

A ROS 2 package for performing object detection on the Renesas RZ/V2H platform. Implements various detection models and provides nodes for both static image and camera-based detection.

Overview

This package provides ROS 2 nodes for:

• General object detection using YOLOX Pascal VOC model.

• Hand detection using YOLOX and Gold YOLOX models.

• Rock-Paper-Scissors hand detection using YOLOv8 models.

• Real-time camera-based detection.

• Static image-based detection.

Features

• Support for multiple detection models:

  • YOLOX Pascal VOC (20 classes)

  • YOLOX Hand Detection

  • Gold YOLO Hand Detection

  • YOLOv8 RPS Hand Detection

• Configurable detection parameters.

• Integration with Foxglove Studio for visualization.

• Multi-threaded processing support.

Node Parameters

Parameter Name	Data Type	Default	Description
model_path	string	“”	Path to the model file.
model_type	string	“yolox_pascal_voc”	Type of model to use (yolox_pascal_voc, yolox_hand, gold_yolox_hand, yolov8_rps).
processing_queue_size	int	5	Size of the image processing queue.
confidence_threshold	float	0.5	Threshold for detection confidence.
iou_threshold	float	0.45	Threshold for IoU in NMS.
class_names	string[]	[]	Optional custom class names.
processing_threads	int	1	Number of processing threads.

Topics

Subscriptions

Topic	Type	Description
/image_raw	sensor_msgs/msg/Image	Input images for object detection.

Publishers

Topic	Type	Description
bounding_box	geometry_msgs/msg/PoseArray	Detection results as pose arrays.
object_detect	std_msgs/msg/String	Detected object name.

Launch Files

Camera-based Detection

# Hand detection using camera input
ros2 launch rzv_object_detection camera_hand_detection_yolox.launch.py

# RPS hand detection using camera input
ros2 launch rzv_object_detection camera_rps_hand_detection_yolov8.launch.py

Static Image Detection

# Pascal VOC object detection on static image using YOLOX
ros2 launch rzv_object_detection static_object_detection_yolox.launch.py

# Hand detection on static image using YOLOX
ros2 launch rzv_object_detection static_hand_detection_yolox.launch.py

# Hand detection on static image using Gold YOLO
ros2 launch rzv_object_detection static_hand_detection_gold_yolo.launch.py

# RPS hand detection on static image using YOLOv8
ros2 launch rzv_object_detection static_rps_hand_detection_yolov8.launch.py

Usage Examples

# Run with custom model type and thresholds
ros2 run rzv_object_detection yolox_object_detection \
    --ros-args -p model_type:=yolox_hand \
    -p confidence_threshold:=0.6 -p iou_threshold:=0.4

# Run with specific model path
ros2 run rzv_object_detection yolox_object_detection \
    --ros-args -p model_path:=/path/to/custom/drpai/model

# Launch with custom camera device
ros2 launch rzv_object_detection camera_hand_detection_yolox.launch.py \
    video_device:=/dev/video1

Dependencies

• ROS 2

• OpenCV

• rzv_model_utils_ros2

• rzv_model and related AI model packages

• image_publisher (for static images)

• v4l2_camera (for camera input)

• foxglove_keypoint_publisher (for visualization)

License

This package is licensed under Apache 2.0.