Large-Scale Security Seal Verification with Computer Vision Algorithms

Nowadays, counterfeit prevention is becoming a crucial challenge in our world-wide globalized market, particularly for high-value and premium brands, whose products are increasingly easy to replicate due to advances in manufacturing technologies and mass production. Ensuring authenticity in such contexts is critical to protect both brand integrity and consumer trust. In this paper, we present a scalable, computer vision-based system for verifying security seals in real-world conditions with the use of a smartphone. The proposed solution builds on top of custom adhesive labels, specifically engineered to embed fibres that act as unique fingerprints. A dedicated recognition pipeline, co-designed in parallel with the label, combines multiple visual inspection modules to assess authenticity in an automatic and robust way. The proposed system, currently deployed in a sandbox end-to-end web application for testing, operates without the need of expert input, and will be soon ready for large-scale deployment.