Driver Activity Classification Using Generalizable Representations from Vision-Language Models

TL;DR

This paper introduces a novel driver activity classification method using vision-language models, achieving high accuracy and robustness by leveraging contrastively-learned representations and a specialized neural network for multi-perspective video analysis.

Contribution

The paper presents SRLF-Net, a new neural network architecture that fuses vision-language embeddings for improved driver activity recognition across diverse scenarios.

Findings

Achieves strong accuracy on the Naturalistic Driving Action Recognition Dataset.

Demonstrates robustness and interpretability through natural language descriptors.

Leverages contrastively-learned vision-language representations for generalization.

Abstract

Driver activity classification is crucial for ensuring road safety, with applications ranging from driver assistance systems to autonomous vehicle control transitions. In this paper, we present a novel approach leveraging generalizable representations from vision-language models for driver activity classification. Our method employs a Semantic Representation Late Fusion Neural Network (SRLF-Net) to process synchronized video frames from multiple perspectives. Each frame is encoded using a pretrained vision-language encoder, and the resulting embeddings are fused to generate class probability predictions. By leveraging contrastively-learned vision-language representations, our approach achieves robust performance across diverse driver activities. We evaluate our method on the Naturalistic Driving Action Recognition Dataset, demonstrating strong accuracy across many classes. Our results suggest that vision-language representations offer a promising avenue for driver monitoring systems, providing both accuracy and interpretability through natural language descriptors.