jsk_recognition

The stack is organized into focused packages like jsk_perception for 2D images and jsk_pcl_ros for point clouds, allowing selective use and easier integration into existing ROS pipelines, as shown in the README's package table.

Each package has dedicated documentation links and build status badges for multiple ROS distributions, indicating good maintenance and support for setups like Kinetic, Melodic, and Noetic.

Developed by the JSK lab, it's used in academic and research robotics, providing confidence in its robustness for real-world perception tasks, as noted in the project philosophy.

Includes ready-to-use nodes for SIFT feature extraction and checkerboard detection, saving development time for common computer vision tasks like camera calibration and feature matching.