An Enhanced Intelligent Transport Mechanism for Detecting and Recognizing Traffic Signs using Transformer with YOLOv9 Model

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Traffic road sign detection and classification are essential technologies for developing an automated driver guidance system using computer-aided techniques. Accidents are increasing on a daily basis because of improper and irregular road signs. Many studies have already been reported on the identification, recognition, and classification of traffic signs on roads. These studies have limitations such as long recognition times, misidentification, fewer results, and the absence of image enhancement. To address these issues, this paper proposes an enhanced intelligent transport mechanism for detecting and recognizing traffic signs using a transformer and a YOLOv9 model. Initially, images are collected from publicly available datasets, and then geometric and photometric distortion approaches are used to augment the increasing number of images and reduce overfitting issues. The images are preprocessed using the bilateral filter to remove artifacts present in the images. The global feature modeling transformer-based encoder-decoder model is used to enhance images and remove unwanted artifacts. Lastly, a refined YOLOv9 model is proposed to identify and detect traffic signs in a variety of weather scenarios. Performance is improved and adjusted using the updated wild horse optimizer. On the Indian traffic sign dataset, German traffic sign recognition benchmark, and Belgian traffic sign dataset, the proposed model achieves accuracy of 99.52%, 99.81%, and 99.39%, respectively.