Tyre Defect Detection using Deep Learning Technique

Abstract

Tyre defects can pose safety hazards and increase maintenance costs. To address this issue, tyre defect detection using digital image processing and convolutional neural networks (CNN) has gained increasing attention recently. Given many manufactured items, the final quality inspection of mass-produced tyres is challenging. Quality aspects include materials, geometry, appearance and final functionality of the tyre. Visually, tyres are characterized by formal features such as annotations and barcodes, which are necessary for identifying the product. A high-quality product has a visually defect-free appearance. Digitizing the final quality inspection before the product leaves the factory is crucial to ensuring high quality. The industrial revolution strongly emphasizes automation, machine learning, sensory systems, digitization and data visualization, all of which are part of Industry 4.0 processes. Visual inspection of tyres is crucial to safe driving during and after the quality grading process. This study proposes a novel modified deep learning technique for inspecting tyres, achieving an average precision of 82.39% and detecting tyres at an average of 1.158 seconds, making them ideal for industrial use.