Counterfeit products caught out by machine-learning tool
Image credit: Dreamstime
Researchers at New York University have developed machine-learning tools which analyse detailed photographs to identify counterfeit handbags, electronics and other products with an accuracy of approximately 98 per cent.
Nearly every high-value object – from watches to helicopters – will suffer from the manufacture of counterfeits.
The growth of the market for counterfeit consumer products has proved a major problem all around the world. Researchers have found that counterfeit trafficking has increased by 10,000 per cent in the past 20 years and now makes up 5-7 per cent of the world’s trade. Profits collected by counterfeit traders have been shown to be a major source of funding for activities such as human trafficking and terrorism.
Existing methods to detect counterfeit products tend to be invasive and risk damaging high-value products. Researchers at New York University have now developed a new, non-invasive detection method which only requires a high-quality photograph of the product.
“The underlying principle of our system stems from the idea that microscopic characteristics in a genuine product or a class of products – corresponding to the same larger product line – exhibit inherent similarities that can be used to distinguish these products from their corresponding counterfeit versions,” said Professor Lakshminarayanan Subramanian, who leads the Open Networks and Big Data Lab at New York University.
The hardware necessary for their detection method is a wide-angle microscopy device, compatible with a smartphone. This allows the user to capture a microscopic image of the object being tested. Based on these microscopic images, supervised machine-learning algorithms – trained on a dataset of known genuine and counterfeit products – are able to classify the product.
According to the researchers, the use of microscopic photographs allows for the detection of “‘super-fake’ counterfeits observed in the marketplace that are not easily discernible [by] the human eye.”
Professor Subramanian and his team put the system to the test with images of fabric, pills, leather, electronics, toys and shoes. They found that the system returned accurate results approximately 98 per cent of the time.
The counterfeit detection system is being brought to market by Entrupy Inc, a start-up founded by Professor Subramanian and colleagues. So far, Entrupy has assessed $14 million of products for authenticity.