SAMIRO: Spatial Attention Mutual Information Regularization with a Pre-trained Model as Oracle for Lane Detection

2025-11-13

AI Summary

Real-world lane detection faces challenges including cluttered backgrounds, illumination variations, and occlusions, while existing data-driven methods suffer from limited generalization. To address this, we propose SAMIRO—a novel framework that jointly leverages spatial attention and mutual information regularization, guided by a pre-trained model serving as an oracle for cross-architectural, plug-and-play knowledge transfer. SAMIRO explicitly models long-range spatial dependencies, preserves domain-agnostic structural priors, and enforces mutual information maximization over feature distributions to suppress background interference. Evaluated on CULane, TuSimple, and LLAMAS benchmarks, SAMIRO consistently improves the performance of diverse mainstream detectors, significantly enhancing robustness and generalization across varying conditions. These results validate its universality and practical applicability in real-world lane detection.

Technology Category

• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Lane detection is an important topic in the future mobility solutions. Real-world environmental challenges such as background clutter, varying illumination, and occlusions pose significant obstacles to effective lane detection, particularly when relying on data-driven approaches that require substantial effort and cost for data collection and annotation. To address these issues, lane detection methods must leverage contextual and global information from surrounding lanes and objects. In this paper, we propose a Spatial Attention Mutual Information Regularization with a pre-trained model as an Oracle, called SAMIRO. SAMIRO enhances lane detection performance by transferring knowledge from a pretrained model while preserving domain-agnostic spatial information. Leveraging SAMIRO's plug-and-play characteristic, we integrate it into various state-of-the-art lane detection approaches and conduct extensive experiments on major benchmarks such as CULane, Tusimple, and LLAMAS. The results demonstrate that SAMIRO consistently improves performance across different models and datasets. The code will be made available upon publication.

Problem

• Addressing lane detection challenges in varying illumination and occlusions
• Reducing data collection and annotation costs for lane detection
• Enhancing contextual and global information utilization in lane detection

Innovation

• Spatial attention mutual information regularization for lane detection
• Knowledge transfer from pretrained model as oracle
• Plug-and-play integration with existing detection methods