A Transfer Learning Approach for Tyre Wear Classification

Automobiles have become an indispensable mode of transportation and travel as the economy has grown. Tires are vital components of automobiles, and their wear contributes to a huge number of road accidents. As a result, anticipating tyre life has become one of the most important criteria to avoid accidents. The proposed work classifies the tire wear condition such as No wear, Light Wear, Moderate Wear, High Wear, and Extreme Wear using a pre-trained deep learning model. The first tyre dataset was created with 1403 total images in 5 classes. Then the tyre dataset was used to classify the tire wear condition by the pre-trained model ResNet50. When compared to other pre-trained models such as SqueezNet, ResNet18, ResNet101, and InceptionResNet, the ResNet50 provided better recognition accuracy of 95.1%. The performance of the proposed method was compared with state of art methods.