MILAN: Milli-Annotations for Lidar Semantic Segmentation

TL;DR

This paper introduces MILAN, a method that uses self-supervised lidar representations to significantly reduce annotation costs for lidar semantic segmentation by selecting informative scans and clustering points for minimal labeling.

Contribution

MILAN demonstrates that self-supervised lidar representations enable effective scan selection and clustering, drastically reducing annotation effort while maintaining high segmentation performance.

Findings

Scan selection with self-supervised features outperforms random selection.

Clustering-based annotation achieves near full annotation performance.

Requires only one thousandth of point labels for effective training.

Abstract

Annotating lidar point clouds for autonomous driving is a notoriously expensive and time-consuming task. In this work, we show that the quality of recent self-supervised lidar scan representations allows a great reduction of the annotation cost. Our method has two main steps. First, we show that self-supervised representations allow a simple and direct selection of highly informative lidar scans to annotate: training a network on these selected scans leads to much better results than a random selection of scans and, more interestingly, to results on par with selections made by SOTA active learning methods. In a second step, we leverage the same self-supervised representations to cluster points in our selected scans. Asking the annotator to classify each cluster, with a single click per cluster, then permits us to close the gap with fully-annotated training sets, while only requiring one thousandth of the point labels.