Deep Learning-Based Object Detection for Autonomous Vehicles: A Comparative Study of One-Stage and Two-Stage Detectors on Basic Traffic Objects

TL;DR

This study directly compares one-stage (YOLOv5) and two-stage (Faster R-CNN) object detectors for autonomous driving using a balanced real+synthetic dataset of Cars, Pedestrians, and Trucks drawn from BDD100K and SHIFT. It evaluates accuracy (COCO-style metrics $mAP_{50-95}$, $mAP_{50}$, recall) and speed (FPS) across varying dataset sizes, resolutions, and lighting/traffic conditions, revealing that YOLOv5 excels in overall accuracy and training efficiency as data grows, while Faster R-CNN shows advantages for small/distant objects and challenging lighting. The results inform deployment guidance: use YOLOv5 for real-time requirements and large-scale data, Faster R-CNN for robustness to occlusion and low-light, and consider hybrid cascades to leverage the strengths of both. Limitations include not assessing newer architectures like YOLOv8 or DETR, and focusing on three object classes with static imagery, suggesting future work to broaden conditions and sensor modalities.

Abstract

Object detection is a crucial component in autonomous vehicle systems. It enables the vehicle to perceive and understand its environment by identifying and locating various objects around it. By utilizing advanced imaging and deep learning techniques, autonomous vehicle systems can rapidly and accurately identify objects based on their features. Different deep learning methods vary in their ability to accurately detect and classify objects in autonomous vehicle systems. Selecting the appropriate method significantly impacts system performance, robustness, and efficiency in real-world driving scenarios. While several generic deep learning architectures like YOLO, SSD, and Faster R-CNN have been proposed, guidance on their suitability for specific autonomous driving applications is often limited. The choice of method affects detection accuracy, processing speed, environmental robustness, sensor integration, scalability, and edge case handling. This study provides a comprehensive experimental analysis comparing two prominent object detection models: YOLOv5 (a one-stage detector) and Faster R-CNN (a two-stage detector). Their performance is evaluated on a diverse dataset combining real and synthetic images, considering various metrics including mean Average Precision (mAP), recall, and inference speed. The findings reveal that YOLOv5 demonstrates superior performance in terms of mAP, recall, and training efficiency, particularly as dataset size and image resolution increase. However, Faster R-CNN shows advantages in detecting small, distant objects and performs well in challenging lighting conditions. The models' behavior is also analyzed under different confidence thresholds and in various real-world scenarios, providing insights into their applicability for autonomous driving systems.

Figures (7)

• Figure 1: Overview of the different annotation types in the BDD100K dataset yu2020bdd100k
• Figure 2: Overview of the different annotation types in the SHIFT dataset sun2022shift
• Figure 3: Comparison of mAP@50 and mAP@50-95 between YOLOv5 and Faster R-CNN
• Figure 4: Scenario 1: High traffic density,
• Figure 5: Scenario 2: low traffic density