Automated evaluation of anti-counterfeiting measures

Patent metadata

Field	Value
Publication number	US-12131335-B2
Application number	US-202017028897-A
Country	US
Kind code	B2
Filing date	Sep 22, 2020
Priority date	Sep 22, 2020
Publication date	Oct 29, 2024
Grant date	Oct 29, 2024

Abstract

A system for evaluating the effectiveness of an anti-counterfeiting measure employed for an item is provided. The system trains a classifier to indicate whether the anti-counterfeiting measure of an evaluation item is genuine or counterfeit. For evaluation items that have been classified as genuine or counterfeit, the system applies the classifier to determine whether the anti-counterfeiting measure of that evaluation item is genuine or counterfeit. The system then generates an effectiveness metric that indicates whether the anti-counterfeiting measure is effective based on evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit.

First claim

We claim:

1. A method performed by one or more computing systems for evaluating the effectiveness of an anti-counterfeiting measure employed for a first item, the method comprising:
accessing one or more evaluation features derived from evaluation items that are assessed as genuine or counterfeit;
for each evaluation item, applying an evaluation classifier to the one or more evaluation features to generate an indication of whether the anti-counterfeiting measure of the evaluation item is counterfeit, the classifier having been trained using training data that includes one or more training features of training items that are labeled based on whether the anti-counterfeiting measure of the training item is genuine;
counting a number of evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit, to produce a count of evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit;
generating an effectiveness metric as an indication of whether the anti-counterfeiting measure is effective, based on the count of evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit;
making a determination of whether a second item is counterfeit based on an assessment of the second item and the effectiveness metric, wherein the assessment of the second item includes processing one or more input features of the second item with an autoencoder to generate one or more output features, the autoencoder having been trained using training data that includes the one or more input features from training items that are genuine, to set weights of the autoencoder;
generating a difference map that indicates differences at a pixel level between the one or more input features and the one or more output features; and
generating an authentication metric based on the difference map; and
outputting an indication of the determination of whether the second item is counterfeit based on the authentication metric and the effectiveness metric.

2. The method of claim 1 wherein the classifier is a convolutional neural network.

3. The method of claim 1 wherein the evaluation items are assessed as being genuine or counterfeit based on an assessment system.

4. The method of claim 3 wherein the effectiveness of the anti-counterfeiting measure is specific to the assessment system.

5. The method of claim 1 wherein the one or more evaluation features include an image of at least a portion of the evaluation item.

6. The method of claim 1 wherein the one or more evaluation features include a non-image feature derived from the evaluation item.

7. The method of claim 1 further comprising training the classifier using training data that includes one or more features of training items that are labeled based on whether the anti-counterfeiting measure of the training items is genuine.

8. One or more computing systems for evaluating the effectiveness of an anti-counterfeiting measure employed for a first item, the one or more computing systems comprising:
one or more computer-readable storage mediums for storing computer-executable instructions for controlling the one or more computing systems to:
for each of a plurality of evaluation items,
derive a feature from the evaluation item; and
apply a classifier to the evaluation feature to generate a classification as to whether the anti-counterfeiting measure of the evaluation item is counterfeit, the classifier having been trained using training data that includes, for each training item, a feature derived from that training item and a label indicating whether the anti-counterfeiting measure of the training item is genuine;

count a number of evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit, to produce a count of evaluation items that are assessed as being genuine whose anti-counterfeiting measures are classified as being counterfeit;
output an indication of whether the anti-counterfeiting measure is effective based on the count of evaluation items that are assessed as being genuine and whose anti- counterfeiting measures are classified as counterfeit;
make a determination of whether a second item is counterfeit based on an assessment of the second item and an effectiveness metric, wherein the assessment of the second item includes processing one or more input features of the second item with an autoencoder to generate one or more output features, the autoencoder having been trained using training data that includes the one or more input features from training items that are genuine, to set weights of the autoencoder;
generating a difference map that indicates differences at a pixel level between the one or more input features and the one or more output features; and
generating an authentication metric based on the difference map; and
output an indication of the determination of whether the second item is counterfeit based on the authentication metric and the effectiveness metric; and
one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.

9. The one or more computing systems of claim 8 wherein the instructions further control the one or more computing systems to train the classifier.

Assignees

• L Livermore Nat Security Llc

Inventors

• Giera Brian
• Krauter Kerry G
• Wihl Brian Matthew
• Zelinski Michael E

Classifications

• G06N3/0475

  Generative networks · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06V20/20

  in augmented reality scenes · CPC title