BaSAL: Size-Balanced Warm Start Active Learning for LiDAR Semantic Segmentation

arXiv:2310.08035v2 Announce Type: replace
Abstract: Active learning strives to reduce the need for costly data annotation, by repeatedly querying an annotator to label the most informative samples from a pool of unlabeled data, and then training a model from these samples. We identify two problems with existing active learning methods for LiDAR semantic segmentation. First, they overlook the severe class imbalance inherent in LiDAR semantic segmentation datasets. Second, to bootstrap the active learning loop when there is no labeled data …

active learning arxiv cs.cv lidar segmentation semantic type warm