Spatial Detection Network using ROS2 Humble

OscarAmes

HI. I am attempting to use an oak-d-lite for object detection and localization. The idea is to identify objects using a yolo like model and then use the depth information to calculate the position of the object relative to the camera. I would like to output both the positions, rgb image and bboxes on said rgb image. this will be processed by a respbery pi or a jetson xavier running ros2 humble.

i have found the option for the Spatial Detection Network which appears to do what i want to achieve however i am having may issues attempting to use the depthairosdriver to achieve this. is it possible to just use depthai as a library and not utilize the driver or is there a method through the driver to achieve the desired result.

link to spacial detection network:

https://docs.luxonis.com/software-v3/depthai/examples/spatial_detection_network/spatial_detection/#Spatial%20Detection%20Network

jakaskerl

OscarAmes cc @Luxonis-Adam

Luxonis-Adam

Hi, what issues are you experiencing?
As an alternative to depthai_ros_driver you can use stereo_inertial_node in depthai_examples which has a similar pipeline, although it has a fixed pipeline. You can also use it as a template for creating your own executable which you can implement in your own package.