Advances in Lidar Point Cloud Segmentation for automotive applications : segmenting outside the box / von Frederik Lenard Hasecke. Wuppertal, 2023

Inhalt

2 Fundamentals

2.1 Lidar Sensor Fundamentals

2.1.1 Automotive Lidar Sensors

2.2 Deep Learning

2.3 Deep Learning on Lidar Data

I Developing Novel Algorithms and Networks for Lidar Segmentation

3 Contributions to Instance Segmentation: Developing a New Clustering Algorithm for Lidar Data

3.1 Related Work

3.1.1 Clustering Algorithms

3.2 Fast Lidar Image Clustering: Method

3.3 Fast Lidar Image Clustering: Evaluation

3.4 Conclusion

4 Contributions to Semantic Segmentation: Creating an Advanced Network Architecture for Three-Dimensional Semantic Segmentation of Lidar Point Clouds

4.3 RangePillars: Evaluation

4.4 Conclusion

5 Contributions to Panoptic Segmentation: Developing Novel and Improved Methods for Panoptic Point Cloud Segmentation

5.3 Lidar Image Panoptic Segmentation

5.4 Conclusion

II Enhancing Lidar Segmentation Perception and Adaptability: Novel Techniques for Data Augmentation and Domain Adaptation

6 Developing Advanced Lidar Point Cloud Augmentation Methods for Improved Segmentation

6.1 Related Work

6.2 Structure Aware Lidar Augmentation Methods

6.3 Evaluation

6.4 Conclusion

7 Enhancing Lidar Domain Adaptation for Robust Semantic Segmentation

7.3 Evaluation

7.4 Conclusion

III Expanding the Horizons of Autonomous Driving: Novel Applications of Lidar Segmentation