ROS2 NanoOWL for Open-vocabulary object detection

Presenting ROS2 NanoOWL - a ROS 2 node for open-vocabulary object detection using NanoOWL. NanoOWL optimizes OWL-ViT to run real-time on NVIDIA Jetson Orin using TensorRT.



  • Takes image input from a rosbag, Isaac Sim or live camera stream.
  • Takes input queries from the user as a list of objects via command line or Foxglove publish panel. This query can be changed anytime while the node is running to detect different objects.
  • Visualizes results as bounding boxes and labels on RViz or Foxglove.

Check out our Isaac SIM Jetson HIL Course Doc to learn how to use ROS2 NanoOWL with Isaac Sim. This course has information on setting up Isaac Sim, running Isaac ROS on NVIDIA Jetson and testing a simulated robot with Hardware-in-the-loop.

Visit ROS2 NanoOWL on GitHub to learn more!

1 Like