LRPD: Long Range 3D Pedestrian Detection Leveraging Specific Strengths of LiDAR and RGB

While short range 3D pedestrian detection is sufficient for emergency\nbreaking, long range detections are required for smooth breaking and gaining\ntrust in autonomous vehicles. The current state-of-the-art on the KITTI\nbenchmark performs suboptimal in detecting the position of pedestrians at long\nrange. Thus, we propose an approach specifically targeting long range 3D\npedestrian detection (LRPD), leveraging the density of RGB and the precision of\nLiDAR. Therefore, for proposals, RGB instance segmentation and LiDAR point\nbased proposal generation are combined, followed by a second stage using both\nsensor modalities symmetrically. This leads to a significant improvement in mAP\non long range compared to the current state-of-the art. The evaluation of our\nLRPD approach was done on the pedestrians from the KITTI benchmark.\n