Cross-View Cross-Modal Unsupervised Domain Adaptation for Driver Monitoring System

TL;DR

This paper introduces a two-phase unsupervised domain adaptation framework that jointly addresses cross-view and cross-modal challenges in driver monitoring, significantly improving activity recognition accuracy across diverse vehicle setups.

Contribution

The work presents a novel joint framework combining contrastive learning and information bottleneck for cross-view and cross-modal unsupervised domain adaptation in driver monitoring.

Findings

Improves top-1 accuracy on RGB video data by nearly 50%.

Outperforms existing unsupervised domain adaptation methods by up to 5%.

Effectively handles real-time driver activity recognition across diverse vehicle configurations.

Abstract

Driver distraction remains a leading cause of road traffic accidents, contributing to thousands of fatalities annually across the globe. While deep learning-based driver activity recognition methods have shown promise in detecting such distractions, their effectiveness in real-world deployments is hindered by two critical challenges: variations in camera viewpoints (cross-view) and domain shifts such as change in sensor modality or environment. Existing methods typically address either cross-view generalization or unsupervised domain adaptation in isolation, leaving a gap in the robust and scalable deployment of models across diverse vehicle configurations. In this work, we propose a novel two-phase cross-view, cross-modal unsupervised domain adaptation framework that addresses these challenges jointly on real-time driver monitoring data. In the first phase, we learn view-invariant and action-discriminative features within a single modality using contrastive learning on multi-view data. In the second phase, we perform domain adaptation to a new modality using information bottleneck loss without requiring any labeled data from the new domain. We evaluate our approach using state-of-the art video transformers (Video Swin, MViT) and multi modal driver activity dataset called Drive&Act, demonstrating that our joint framework improves top-1 accuracy on RGB video data by almost 50% compared to a supervised contrastive learning-based cross-view method, and outperforms unsupervised domain adaptation-only methods by up to 5%, using the same video transformer backbone.