The analysis of acquisition system for electronic traffic signal in smart cities based on the internet of things
Abstract This work designs an intelligent traffic electronic information signal acquisition system based on the Internet of Things (IoT) and deep learning (DL). It aims to address the increasingly severe congestion issues in urban traffic and improve the efficiency and intelligence of traffic manage...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-07-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-07423-6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract This work designs an intelligent traffic electronic information signal acquisition system based on the Internet of Things (IoT) and deep learning (DL). It aims to address the increasingly severe congestion issues in urban traffic and improve the efficiency and intelligence of traffic management. First, a system framework is constructed that includes three core modules: video acquisition and transmission (VAT), video processing (VP), and information processing. The system captures real-time video information of traffic scenes through cameras. Meanwhile, vehicle detection and tracking are performed using a video image processor (VIP) to extract traffic parameters, which are then transmitted to the traffic control platform. Second, an improved Multi-Task Convolutional Neural Network (MT-CNN) model, called Attention-Mechanism Multi-Modal Feature Fusion GooGleNet (AM-MMFF-GooGleNet), is proposed. This model integrates Multi-Modal Feature Fusion (MMFF) and Channel Attention Mechanism (AM), significantly improving the accuracy and robustness of vehicle localization and identification. Experimental results show that the AM-MMFF-GooGleNet model achieves an accuracy of 98.6% in the vehicle localization task, 3.2% higher than the original MT-GooGleNet. The accuracy under different lighting conditions and high background noise scenarios reaches 97.3%, 96.8%, and 95.5%, demonstrating strong environmental adaptability. Furthermore, the average detection time of the model is 20.5 milliseconds, indicating good real-time performance. By optimizing the DL model and system design, the ability to acquire and process vehicle electronic information signals in the intelligent transportation system (ITS) is remarkably enhanced, providing more precise decision-making support for traffic management. This work offers an innovative technical solution for the development of ITS, promoting the deep integration and application of IoT and DL technologies in the traffic field. Thus, the work provides strong technical support for alleviating urban traffic congestion and improving traffic efficiency. |
|---|---|
| ISSN: | 2045-2322 |