System and method for handbag authentication
Systems and methods for authenticating handbags using a portable electronic device along with a bilinear convolutional neural network (CNN) model are described. One method includes using a portable electronic device comprising a camera, and a lens-accessory attached to the portable electronic device such that an optical feature of the lens-accessory is positioned in front of the camera. The portable electronic device acquires one or more pictures of a handbag and sends the one or more pictures to a bilinear CNN model via a network asset where an authenticity is determined. The systems and methods disclosed are capable of allowing the portable electronic device to be spaced apart from the handbag while acquiring pictures, and the lens-accessory can be between 10× and 50× magnification.
1. A method of authenticating a handbag comprising:
receiving a picture of the handbag having a plurality of authentic-identifying features;
receiving one or more characteristics of the handbag;
determining a fabric associated with the handbag based on the one or more characteristics;
selecting from among a plurality of counterfeit-classifier models each trained on one type of fabric, a counterfeit-classifier model trained on the fabric associated with the handbag;
processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag; and
determining authentication of said handbag.
2. The method of claim 1 , wherein the one or more characteristics comprise brand, style, the fabric, or a combination thereof.
3. The method of claim 1 , wherein the counterfeit-classifier model comprises a bilinear convolutional neural network (CNN) model trained from plurality of images.
4. The method of claim 1 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag comprises:
generating a first output from a first feature extracted from a first feature extractor;
generating a second output from a second feature extracted from a second feature extractor; and
multiplying the first and second outputs by an outer product.
5. The method of claim 1 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with handbag comprises:
extracting the plurality of authentic-identifying features of the pictures;
comparing the plurality of authentic-identifying features of the picture with a plurality of authentic-identifying features associated with the counterfeit-classifier model.
6. A system for authenticating a handbag comprising:
a processor and a non-transitory computer-readable medium, the non-transitory computer-readable medium of the electronic device being configured to store instructions, the instructions when executed by the processor cause the processor to perform steps comprising:
receiving a picture of the handbag having a plurality of authentic-identifying features;
receiving one or more characteristics of the handbag;
determining a fabric associated with the handbag based on the one or more characteristics;
selecting from among a plurality of counterfeit-classifier models each trained on one type of fabric, a counterfeit-classifier model trained on the fabric associated with the handbag;
processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag; and
determining authentication of said handbag.
7. The system of claim 6 , wherein the one or more characteristics comprise brand, style, the fabric, or a combination thereof.
8. The system of claim 6 , wherein the counterfeit-classifier model comprises a bilinear convolutional neural network (CNN) model trained from plurality of images.
9. The system of claim 6 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag comprises:
generating a first output from a first feature extracted from a first feature extractor;
generating a second output from a second feature extracted from a second feature extractor; and
multiplying the first and second outputs by an outer product.
10. The system of claim 6 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with handbag comprises: extracting the plurality of authentic-identifying features of the pictures; comparing the plurality of authentic-identifying features of the picture with a plurality of authentic-identifying features associated with the counterfeit-classifier model.
11. A non-transitory computer-readable medium configured to store instructions, the instructions when executed by one or more computers, cause the one or more computers to perform operations comprising:
receiving a picture of a handbag having a plurality of authentic-identifying features;
receiving one or more characteristics of the handbag;
determining a fabric associated with the handbag based on the one or more characteristics;
selecting from among a plurality of counterfeit-classifier models each trained on one type of fabric, a counterfeit-classifier model trained on the fabric associated with the handbag;
processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag; and
determining authentication of said handbag.
12. The non-transitory computer-readable medium of claim 11 , wherein the one or more characteristics comprise brand, style, the fabric, or a combination thereof.
13. The non-transitory computer-readable medium of claim 11 , wherein the counterfeit-classifier model comprises a bilinear convolutional neural network (CNN) model trained from plurality of images.
14. The non-transitory computer-readable medium of claim 11 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with the handbag comprises:
generating a first output from a first feature extracted from a first feature extractor;
generating a second output from a second feature extracted from a second feature extractor; and
multiplying the first and second outputs by an outer product.
15. The non-transitory computer-readable medium of claim 11 , wherein the processing the picture through the counterfeit-classifier model trained on the fabric associated with handbag comprises:
extracting the plurality of authentic-identifying features of the pictures;
comparing the plurality of authentic-identifying features of the picture with a plurality of authentic-identifying features associated with the counterfeit-classifier model.