Deep Learning-Based Tyre Quality Classification Using ResNet50: A Robust Approach for Automated Inspection and Defect Detection

The classification of tyre quality in the present work is into two groups, namely, defective and good condition, by using a deep learning model called ResNet50. It intends to leverage the power of CNNs so that tyre defect detection can be fully automated and done with higher accuracy, which is very important with regard to safety, performance, and efficiency within the automotive sector. It includes 1,854 high-resolution images of tyres from Kaggle, with a split of 80/20 between the training and validation sets. Preprocessing techniques include image resizing, normalizing, and augmenting to enhance model training. Early stopping at 10 epochs was done for fine-tuning ResNet50 in order to avoid overfitting as a result of the residual learning structure. An overall accuracy of 95% was achieved, with precision and recall values of 0.98 and 0.94, respectively, for defective tyres, and 0.91 and 0.97 for good tyres. These results were then verified using the confusion matrix and detailed classification report. The study highlights the potential of ResNet50 in classifying tyre quality and presents an appropriate automated inspection system. In the future, further research on other advanced deep learning models may be investigated, more data acquired, and there's a possibility of designing real-time optimized systems for industrial applications to take this tyre inspection process further.