YOLO-TS: Real-Time Traffic Sign Detection with Enhanced Accuracy Using Optimized Receptive Fields and Anchor-Free Fusion

Junzhou Chen,Heqiang Huang,Ronghui Zhang,Nengchao Lyu,Yanyong Guo,Hong-Ning Dai,Hong Yan

2024-10-23

Abstract:Ensuring safety in both autonomous driving and advanced driver-assistance systems (ADAS) depends critically on the efficient deployment of traffic sign recognition technology. While current methods show effectiveness, they often compromise between speed and accuracy. To address this issue, we present a novel real-time and efficient road sign detection network, YOLO-TS. This network significantly improves performance by optimizing the receptive fields of multi-scale feature maps to align more closely with the size distribution of traffic signs in various datasets. Moreover, our innovative feature-fusion strategy, leveraging the flexibility of Anchor-Free methods, allows for multi-scale object detection on a high-resolution feature map abundant in contextual information, achieving remarkable enhancements in both accuracy and speed. To mitigate the adverse effects of the grid pattern caused by dilated convolutions on the detection of smaller objects, we have devised a unique module that not only mitigates this grid effect but also widens the receptive field to encompass an extensive range of spatial contextual information, thus boosting the efficiency of information usage. Evaluation on challenging public datasets, TT100K and CCTSDB2021, demonstrates that YOLO-TS surpasses existing state-of-the-art methods in terms of both accuracy and speed. The code for our method will be available.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper attempts to address the balance between real-time performance and accuracy in traffic sign recognition technology within autonomous driving and advanced driver assistance systems (ADAS). Current methods, although effective, often compromise between these two aspects. To overcome this challenge, the authors propose a new method called YOLO-TS, aiming to achieve real-time and efficient traffic sign detection.

Specifically, the paper focuses on the following aspects:

1. **Small Object Detection**: Traffic signs have low resolution and limited information in images, making small object detection a significant challenge. The paper improves detection accuracy by optimizing the receptive field of multi-scale feature maps to better match the size distribution of traffic signs.

2. **High-Resolution Feature Maps**: Traditional multi-scale detection methods have limitations. Therefore, the paper proposes an innovative approach using high-resolution feature maps for multi-scale detection, enhancing the richness of contextual information and significantly improving accuracy and speed.

3. **Anti-Grid Effect Module (AGRFM)**: To address the grid effect caused by dilated convolutions, the paper designs a unique module that combines standard convolutions with high dilation rate convolutions, maintaining the continuity of feature maps and improving the efficiency and reliability of small object detection.

4. **Experimental Validation**: The paper conducts extensive experiments on public datasets such as TT100K and CCTSDB2021 to validate the effectiveness of YOLO-TS. The results show that YOLO-TS surpasses existing state-of-the-art methods in both accuracy and speed.

In summary, this paper introduces multiple technological innovations to address the issues of real-time performance and accuracy in traffic sign detection, providing new solutions for the safety and reliability of autonomous driving and advanced driver assistance systems.