Automatic Vehicle Detection using DETR: A Transformer-Based Approach for Navigating Treacherous Roads

2025-02-25

AI Summary

To address the insufficient robustness of automated vehicle detection in complex driving environments—such as poor illumination, diverse road geometries, and heterogeneous vehicle categories—this paper introduces Detection Transformer (DETR) to onboard perception for the first time, proposing the Co-DETR collaborative hybrid assignment training framework. Co-DETR innovatively integrates multiple parallel auxiliary heads with a dynamic label assignment strategy, synergizing CNN-based feature extraction, learnable object queries, bipartite matching, and multi-scale auxiliary supervision. This design significantly enhances convergence and robustness against small objects, occlusions, and low-contrast scenes. On the BadODD benchmark, Co-DETR achieves a 5.3% mAP improvement over YOLOv8 and Faster R-CNN. Moreover, it maintains high accuracy and real-time performance under challenging conditions—including rain, fog, nighttime, and narrow curved roads—demonstrating practical viability for autonomous driving systems.

Technology Category

• Computer Vision: Object Detection & Categorization
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Automatic Vehicle Detection (AVD) in diverse driving environments presents unique challenges due to varying lighting conditions, road types, and vehicle types. Traditional methods, such as YOLO and Faster R-CNN, often struggle to cope with these complexities. As computer vision evolves, combining Convolutional Neural Networks (CNNs) with Transformer-based approaches offers promising opportunities for improving detection accuracy and efficiency. This study is the first to experiment with Detection Transformer (DETR) for automatic vehicle detection in complex and varied settings. We employ a Collaborative Hybrid Assignments Training scheme, Co-DETR, to enhance feature learning and attention mechanisms in DETR. By leveraging versatile label assignment strategies and introducing multiple parallel auxiliary heads, we provide more effective supervision during training and extract positive coordinates to boost training efficiency. Through extensive experiments on DETR variants and YOLO models, conducted using the BadODD dataset, we demonstrate the advantages of our approach. Our method achieves superior results, and improved accuracy in diverse conditions, making it practical for real-world deployment. This work significantly advances autonomous navigation technology and opens new research avenues in object detection for autonomous vehicles. By integrating the strengths of CNNs and Transformers, we highlight the potential of DETR for robust and efficient vehicle detection in challenging driving environments.

Problem

• Detects vehicles in diverse driving environments
• Improves accuracy with DETR and Co-DETR
• Enhances training efficiency using auxiliary heads

Innovation

• Uses DETR for vehicle detection
• Implements Co-DETR training scheme
• Combines CNNs with Transformer technology