ros_deep_learning offers deep learning inference nodes for ROS/ROS2, specifically designed for NVIDIA Jetson Nano, TX1, TX2, Xavier, and Orin devices, leveraging TensorRT for optimized performance. This package integrates DNNs from the jetson-inference library, providing pre-trained networks for image classification, object detection, and semantic segmentation, alongside support for custom user-trained models. It includes camera and video streaming nodes compatible with MIPI CSI, V4L2, RTP/RTSP, WebRTC, video files, and image sequences. The tool supports various ROS distributions, including Melodic, Noetic, Foxy, Galactic, Humble, and Iron, and can be installed via source or pre-built containers, making it a versatile solution for robotics developers.
Best used for
Ideal for developers who need to integrate advanced deep learning capabilities into their ROS/ROS2 robotics projects, deploy AI models on NVIDIA Jetson devices, and process real-time camera and video streams for tasks like object detection or semantic segmentation. Especially valuable for robotics engineers building intelligent autonomous systems.
Common actions