Tire Defect Detection Model Using Machine Learning

Visual inspection of defective tires after manufacturing is critical for saving human lives since defective tires may explode and cause many accidents. Automated tire defect detection is incredibly difficult because of its complex anisotropic multi-texture layers. However, with the advancement of technology, Machine Learning (ML) plays an influential role in defect detection of tiers. Therefore, an ML-based tire defect detection model is proposed in this research. First, we have collected and labeled a novel dataset. Then, Gray Level Co-occurrence Matrix (GLCM) is used for the feature extraction stage. After that, 22 statistical metrics are calculated from the created GLCM matrix. Finally, Support Vector Machine, Artificial Neural Network, k-Nearest Neighbors, Decision Tree, and Logistic Regression are trained, fine-tuned, tested, and compared. The results show that the Artificial Neural Network outperformed the compared classifiers in terms of recall, accuracy, and F1 score. Moreover, the suggested feature extraction techniques differentiate between defected and free-defected tire images. The proposed ML model has achieved 100%, 96%, 97.9%, and 99.34% of recall, precision, F1 score, and accuracy respectively.