lidar_super_resolution
A simulation-based deep learning approach to enhance the resolution of 3D lidar point clouds for ground vehicles.
What is lidar_super_resolution?
Lidar Super-resolution is a deep learning framework that enhances the resolution of sparse 3D lidar point clouds for ground vehicles. It converts point clouds into 2D range images and uses a convolutional neural network to increase their resolution, trained entirely on synthetic data from the CARLA simulator. This approach solves the problem of limited lidar resolution without requiring extensive real-world data collection.
Researchers and engineers working on autonomous vehicles, robotics, or computer vision who need to improve the detail and density of lidar point clouds for better perception and navigation.
It offers a cost-effective, simulation-based training pipeline that avoids the need for large real-world datasets, while providing a practical solution for enhancing lidar resolution that can be validated on actual sensor data.
Overview
Simulation-based Lidar Super-resolution for Ground Vehicles
Use Cases
Best For
- Enhancing low-resolution lidar data (e.g., 16-beam) to higher resolutions (e.g., 64-beam) for autonomous driving
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
