Bridging Safety and Technology: Deep Learning Innovations in Automated Tire Evaluation

Acting as the main point of contact with the road, tires are fundamental for vehicle safety, performance, and fuel economy. Tread wear, cracks, bulges, and punctures—defects that over time could cause accidents if left unnoticed—develop in tires. Early discovery of these flaws not only increases safety but also, by enabling prompt repairs, lowers maintenance costs. This work investigates the use of a deep learning-based image classification system, Inspection V3 model, to automatically evaluate tire quality. Processing a dataset of 1,856 photos comprising 830 excellent tires and 1,328 bad tires, the model employs convolutional neural networks (CNNs) to precisely detect tire states. To maximize the pre-trained model, data augmentation methods and fine-tuning of it were used. Over 100 epochs, a batch size of 16 helped the model to be trained with a test accuracy of 96%. This great degree of accuracy shows that the model can precisely detect tire flaws, so it is a useful instrument for automated tire inspections. Particularly for companies running sizable fleets of vehicles, the deployment of such systems can improve road safety, minimize vehicle downtime, and lower running costs.