Lane-Wise Highway Anomaly Detection

2025-05-05

AI Summary

This work addresses the challenge of lane-level anomaly detection in highway surveillance videos. Methodologically, it proposes a scalable and interpretable multimodal temporal analysis framework that requires no dedicated sensors or explicit road modeling—only raw video streams are used. Lane-level temporal features (e.g., traffic volume, occupancy, heavy-vehicle ratio) are extracted via AI vision models (vehicle detection, counting, and classification), then fused within a multi-branch detection architecture integrating deep learning, rule-based engines, and lightweight machine learning. A key contribution is the construction of the first large-scale, expert-validated lane-level anomaly dataset comprising 73,139 annotated samples, covering four anomaly types—including traffic incidents and sensor-related issues (e.g., camera misalignment). Experiments demonstrate that the proposed method consistently outperforms state-of-the-art approaches across precision, recall, and F1-score, significantly improving both detection accuracy and deployment cost-efficiency in real-world scenarios.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

This paper proposes a scalable and interpretable framework for lane-wise highway traffic anomaly detection, leveraging multi-modal time series data extracted from surveillance cameras. Unlike traditional sensor-dependent methods, our approach uses AI-powered vision models to extract lane-specific features, including vehicle count, occupancy, and truck percentage, without relying on costly hardware or complex road modeling. We introduce a novel dataset containing 73,139 lane-wise samples, annotated with four classes of expert-validated anomalies: three traffic-related anomalies (lane blockage and recovery, foreign object intrusion, and sustained congestion) and one sensor-related anomaly (camera angle shift). Our multi-branch detection system integrates deep learning, rule-based logic, and machine learning to improve robustness and precision. Extensive experiments demonstrate that our framework outperforms state-of-the-art methods in precision, recall, and F1-score, providing a cost-effective and scalable solution for real-world intelligent transportation systems.

Problem

• Detects lane-specific highway anomalies using multi-modal data
• Overcomes sensor dependency with AI-powered vision models
• Introduces a novel dataset with expert-validated anomaly classes

Innovation

• AI-powered vision models for lane-specific features
• Multi-branch system combining deep learning and rule-based logic
• Novel dataset with expert-validated lane-wise anomalies