The new technique to spot counterfeit product uses two dimensional (2D) material tags along with AI driven authentication software.
Archa Harikumar H
New 2D tags, AI authentication may spot counterfeit products more accurately | credits: Paula Piccard
The new technique to spot counterfeit product uses two dimensional (2D) material tags along with Artificial Intelligence driven authentication software. It delivers faster and gives more accurate results even under extreme conditions.
The new counterfeiting technique is called "Deepkey". It was developed by an International team of researchers which was led by the National University of Singapore (NUS). In the scientific journal Matter, the team described their work in the study "Multi-generational Crumpling of 2D materials for Anti-counterfeiting patterns with Deep Learning Authentication".
The 2D material secures tags have randomly generated Physically Unclonable Function (PUF) which is being categorised and validated by deep learning model.
The authentication process for spotting counterfeit products require 3.5 minutes which involves scanning the tags under an electronic microscope to detect the PUF pattern, which is then sent to the AI-driven software for validation.
During the release, Wang Xianion, Assistant Professor at NUS faculty said that, with this they have tackled several bottlenecks that other techniques had encountered and the 2D PUF labels are environmentally stable, easy to read, simple and cheap to build. Thus, the adoption of deep learning accelerated overall authentication and pushing this venture to one step further to practical application.
PUF key based techniques usually gives high encoding capabilities which can be used to produce numerous dissimilar patterns. Even though, it makes the pattern authentication process longer when they are performed in a large database.
Xiaonan detailed that they used deep learning model to precategories PUF patterns into subgroups and thus the search and compare algorithm is done in a much smaller database as it shortens the overall authentication time and now they are on other readout methods which an further shorten the processing time and exploring the idea of securing the tags using block chain which will be useful in transparent tracking of the entire supply chain and quality control process.