UdeerLID+: Integrating LiDAR, Image, and Relative Depth with Semi-Supervised

TL;DR

UdeerLID+ tackles road segmentation in urban environments by fusing LiDAR, relative depth, and image data under a semi-supervised framework. It introduces a two-step data-space adaptation that aligns LiDAR to the 2D image plane via an altitude-difference transformation and leverages Depth Anything for monocular depth cues. The method features a multi-sources encoder-decoder with per-modality auxiliary losses and a Meta Pseudo Labels training regime, achieving a MaxF1 of 97.26 on KITTI and surpassing prior approaches. This work advances autonomous driving perception by delivering robust, cross-modal road detection under diverse conditions and limited labeled data.

Abstract

Road segmentation is a critical task for autonomous driving systems, requiring accurate and robust methods to classify road surfaces from various environmental data. Our work introduces an innovative approach that integrates LiDAR point cloud data, visual image, and relative depth maps derived from images. The integration of multiple data sources in road segmentation presents both opportunities and challenges. One of the primary challenges is the scarcity of large-scale, accurately labeled datasets that are necessary for training robust deep learning models. To address this, we have developed the [UdeerLID+] framework under a semi-supervised learning paradigm. Experiments results on KITTI datasets validate the superior performance.

Figures (1)

• Figure 1: UdeerLID+ Pipline