Edge Detection with ROS Noetic
Modular classical & deep-learning edge detection pipeline, featuring real-time C++ and Python nodes.
This project implements a robust and modular edge detection system using classical image processing methods (Canny, Sobel, Laplacian, Prewitt, Roberts) and the deep learning-based Holistically-Nested Edge Detection (HED) technique. These methods are implemented in Python and C++ for flexibility. All nodes publish overlays and 3-D edge markers for RViz, letting you see edges in space, not just in pixels.


Side-by-side: raw image (left) and detected edge overlay (right).
What’s inside
- Edge detectors: Canny (C++ & Python), Sobel, Laplacian, Prewitt, Roberts & HED
- ROS interfaces: launch files for basic tests, and live edge detection in both python and c++ with visualization
- 3-D visualization: publishes 3D overlay and marker
- Reproducible environment: one-command Docker Compose with GPU-ready base image
Live 3-D edge cloud in RViz.