<![CDATA[This research addresses a critical aspect of automotive safety by developing an advanced tire defect classification model to enhance tire maintenance practices and reduce tire-related accidents. The primary objective is to leverage the power of deep learning to accurately distinguish between good and defective or worn-out tires, which is vital for ensuring road safety. The study utilizes a comprehensive dataset encompassing various tire conditions to train and evaluate six prominent deep learning models—EfficientNet, InceptionNet, VGG19, DenseNet121, DenseNet201, and ResNet101—as well as three lightweight models—SqueezeNet, MobileNet V2, and MobileNet V3. Customised Neurawheel models are also introduced and specifically designed for this task. Employing state-of-the-art deep learning and image processing techniques, the models were rigorously trained and tested to ensure high accuracy in classification tasks. Among the models tested, Neurawheel-4j emerges as the top performer, achieving an impressive accuracy rate of 98.44%, significantly outperforming ResNet101 and other models. The research highlights the effectiveness of sophisticated model architectures, rigorous dataset curation, and optimized training configurations, underscoring the potential for these models to be deployed in real-world applications. The implications of this study are profound, as the deployment of such a model in real-world scenarios could dramatically reduce tire-related accidents, contributing to the broader goal of enhancing road safety. Future research should focus on expanding the dataset to include a wider range of real-world scenarios, exploring additional metrics to assess tire wear severity, and integrating the model with IoT-based systems for real-time tire monitoring. This study lays the foundation for further advancements in tire defect classification and automotive safety.]]>
Copyrights © 2025
