Gain more value from your existing Tire Geometry Systems, with the Poling Group's new Tire Visual Inspection System!
Deep Learning for Visual Inspection and Classification of Tire Defects

Deep learning is an artificial intelligence method by which a computer model can parse inputs and produce outputs in a way that is inspired by how neural networks in the human brain work. With the advent of self-driving vehicles, facial recognition, and surveillance cameras able to automatically detect suspicious behavior, computer vision is a quickly developing field within deep learning. Visual identification tasks that were once the sole domains of human inspectors are increasingly achievable by intelligent computer vision systems. Tire manufacturers have an opportunity to use this technology on top of their existing equipment investment to provide even better quality control.

Using images provided by the existing geometry systems, tire manufacturers can prevent tires with visual defects from reaching their customers. Tire factories state that about 1% of tires have curing defects and about 10% of those defects are missed by the human inspectors. Because of these missed defects, factories also report that they re-inspect 5% of their tires.

Our Tire Visual Inspection System uses a combination of rules-based software algorithms, which can provide good results for identifying defects with repeatable features, and Anomaly Detection deep learning models, which are better for high variability and subjective inspection to identify visual defects. With the existing classifiers or secondary inspectors confirming and classifying the potential defects at our Kiosks, an expansive dataset can be created for training an Object Detection deep learning model. This model would continue to receive feedback, even across multiple factories within the same organization, continually re-training itself to be more accurate. This model identifies and classifies defects, allowing those tires to be routed directly to a repair or reject area, without further intervention from the labor force.

Additional Object Detection deep learning models could be trained and implemented to locate and read the DOT Tire Date Code, locate and measure the treadwear indicator bars, and other objects desired by a factory. With the new governmental penalties in place for incorrect DOT Tire Date Codes, reading those codes with our Visual Inspection System can stop tires with incorrect dates from leaving the factory.

Our Tire Visual Inspection System can be scaled and customized to exactly how your factory processes tires. Our models can run on a large server, processing images from multiple machines with tire geometry systems, or they can run directly on the tire geometry systems, using their results to further enhance the tire’s final grade.

For a small investment, a factory can add our Tire Visual Inspection System to detect and classify defects identified in their already-collected tire geometry images. This allows a factory to gain even more value from its prior large investment in tire geometry systems.

Our Tire Visual Inspection System guarantees increased quality from the plant floor to the tires on the road.

Request More Information

Five Divisions with One Goal ; We are committed to being a globally trusted source for tire-testing equipment & software, factory data management, rubber processing equipment, and engineering solutions.

Home Page | Contact | Website Terms & Conditions | Privacy Policy

©2024, the Poling Group. All Rights Reserved.