Context-Aware Sim-to-Real Unsupervised Domain Adaptation for Lane Detection via Disentangled Feature Alignment

While existing supervised deep learning-based lane detection methods achieve exceptional detection performance, constructing a large real-world dataset with labels is a cost-intensive task. Therefore, we propose a novel sim-to-real unsupervised domain adaptation method specialized in lane detection. In this paper, we present a disentangled feature alignment approach that performs selective adaptation for lane and background features. By performing the disentangled prototype-based local and global feature alignments, we solve the negative transfer problem of an existing adversarial feature alignment-based domain adaptation for lane detection. In addition, the lane detection task requires accurate localization of the lanes even in occluded parts. Therefore, we adopt a consistency regularization strategy using masked images to improve the detection accuracy in the occluded area. Additionally, we introduce an adaptive resolution adjustment of the masked-out patch based on training maturity for context learning. By applying a simple framework, we achieve superior lane detection accuracy with lower false positives and false negatives in the target domain compared to conventional methods. Extensive experiments demonstrate the superiority of the proposed method. Note to Practitioners—Lane detection plays a major role in autonomous vehicles by predicting the location of lanes on a road. In particular, deep learning-based lane detection models aim to output robust detection results in various environments, including different weather and road types. Because deep learning models are highly dependent on the distribution of training datasets, models trained with datasets acquired in a given environment can cause significant performance degradation when tested in different environments. Therefore, improving the generalizability of the lane detection model in the real-world application requires a vast training dataset. However, it is impractical to collect and label the real-world datasets including multiple road environments. We propose a novel training strategy for the lane detection models trained with simulation datasets to achieve high detection performance even on the real-world dataset. In other words, we introduce knowledge transfer methods for enhancing stability of the models trained with simulation datasets on real roads. In addition, we expect that the proposed method will transform the manual process of collecting and labeling real-world datasets into automatic creation of simulation-based learning datasets.