We present a surprisingly simple and efficient method for self-supervision of 3D backbone on automotive Lidar point clouds. We design a contrastive loss between features of Lidar scans captured in the same scene. Several such approaches have been proposed in the literature from PointConstrast, which uses a contrast at the level of points, to the state-of-the-art TARL, which uses a contrast at the level of segments, roughly corresponding to objects. While the former enjoys a great simplicity of implementation, it is surpassed by the latter, which however requires a costly pre-processing. In BEVContrast, we define our contrast at the level of 2D cells in the Bird's Eye View plane. Resulting cell-level representations offer a good trade-off between the point-level representations exploited in PointContrast and segment-level representations exploited in TARL: we retain the simplicity of PointContrast (cell representations are cheap to compute) while surpassing the performance of TARL in downstream semantic segmentation.
@article{arxiv.2310.17281,
title = {BEVContrast: Self-Supervision in BEV Space for Automotive Lidar Point Clouds},
author = {Corentin Sautier and Gilles Puy and Alexandre Boulch and Renaud Marlet and Vincent Lepetit},
journal= {arXiv preprint arXiv:2310.17281},
year = {2023}
}