EP-4740122-A2 - ARTWORK AUTHENTICITY SYSTEMS AND METHODS

Abstract

Systems, devices, and methods are disclosed for generating an authenticity score for a creative work using digital image data of one or more creative works. A system receives, via a network, first digital image data of a first creative work. The system determines from the first digital image data, a first set of feature variables, each corresponding to a characteristic of the first creative work. The system determines, via an artificial intelligence model and with input including the first set of feature variables and a comparison dataset, an authenticity score for the first creative work. The artificial intelligence model generates, based on the portion of first feature variables of the first creative work that match corresponding portions of the comparison dataset, an output indicative of an authenticity score. The system communicates, via the network, an authenticity score of the first creative work to a client device.

Inventors

• FRANCESCO, ROCCHI
• FINADRI, ALBERTO

Assignees

• SpaceFarm LLC

Dates

Publication Date

20260513

Application Date

20240702

Claims (20)

1. 1. A system comprising: one or more processors to: receive, via a communication network, a first set of digital images including one or more images of a creative work captured at a first distance; receive first metadata associated with the first set of digital images and including an originality identifier indicative of an originality of the first set of digital images; generate, based on the first set of digital images, one or more image datasets; determine one or more feature variables of the creative work based on the one or more image datasets, each of the one or more feature variables corresponding to a characteristic of the creative work; determine, using a first machine -learning model and according to at least the one or more feature variables of the creative work and a comparison dataset generated from image data for a plurality of separate creative works, an authenticity confidence level of the creative work; and provide, via the communication network and to a client device, the determined authenticity confidence level of the creative work for presentation.
2. 2. The system of claim 1, wherein the one or more processors are further configured to: receive, via the communications network, a second set of digital images and second metadata, wherein the second set of digital images includes one or more images of the creative work captured at a second distance; and generate the one or more image datasets based on the first set of digital images and the second set of digital images.
3. 3. The system of claim 2, wherein the one or more processors are further configured to: receive, via the communications network, a third set of digital images and third metadata, wherein the third set of digital images include one or more images of a backside of the creative work, including two or more images of the entire backside of the creative work; and generate the one or more image datasets based on the first set of digital images, the second set of digital images, and the third set of digital images.
4. 4. The system of claim 3, wherein the third set of digital images comprises one or more images of less than the entire backside of the creative work and one or more images of an author signature.
5. 5. The system of claim 1, wherein the comparison dataset further comprises a plurality of feature variable datasets and a plurality of digital image datasets for each additional creative work of the plurality of additional creative works of the comparison dataset.
6. 6. The system of claim 1, wherein the one or more processors are further configured to: partition, into two or more subsets, each of the following: one more feature variables of the creative work, the one or more image datasets, the one or more datasets extracted from an image dataset, and the first set of digital images; determine an external consistency metric of the two or more subsets based on a comparison of one or more portions of two or more corresponding datasets in each of the different subsets; determine an internal consistency metric for a single subset based on a comparison of one or more portions of data within the same subset; and determine, based on a comparison of the external consistency metric and the internal consistency metric, a confidence score for one or more datasets associated with the digital images of the creative work.
7. 7. The system of claim 1, wherein the one or more processors are further configured to: generate an author profile for a first author of the creative work, the author profile including information for one or more additional creative works of the first author, one or more associated author profiles of one or more authors associated with the first author, one or more time periods, one or more geographic locations, and one or more creative work categories associated with the author; and determine, using the first machine learning model and according to the one or more feature variables of the creative work, the author profile of the first author, and the comparison dataset, the authenticity confidence level of the creative work.
8. 8. The system of claim 7, wherein the one or more processors are further configured to: determine a creative category associated with the creative work; identify a one or more associated creative works associated with the creative category; generate an author profile of a second author based on at least a portion of the comparison dataset that corresponds to one or more creative works authored by the second author and a portion of the comparison dataset that corresponds to the one or more associated creative works; and determine, using the first machine learning model and according to at least the one or more characteristics of the creative work, the author profile for the first author, the author profile for the second author, and the comparison dataset, the authenticity confidence level of the creative work.
9. 9. The system of claim 1, wherein one of the one or more feature variables of the creative work comprises a brush sign frequency map for a subject of the creative work.
10. 10. The system of claim 1, wherein one of the one or more feature variables of the first creative work comprises a pigment density of the first creative work.
11. 11. A method of determining authenticity of a creative work, the method comprising: receiving, via a communications network, a first set of digital images including one or more images of a creative work at a first distance; receiving first metadata associated with the first set of digital images and including an originality identifier indicative of an originality of the first set of digital images; generating, by one or more processors and based on the first set of digital images, a one or more image datasets; determining, by the one or more processors, one or more feature variables of the creative work based on the one or more image data sets, each of the one or more feature variables corresponding to a characteristic of the creative work; determining, using a first machine-learning model and based on at least the one or more feature variables of the creative work and a comparison dataset generated from image data for a plurality of separate creative works, an authenticity confidence level; and providing, via the communication network and to a client device, the determined authenticity confidence level of the creative work for display.
12. 12. The method of claim 11, further comprising: receiving, via the communications network, a second set of digital images and second metadata, wherein the second set of digital images includes one or more images of the creative work at a second distance; and generating, by the one or more processors, the one or more image datasets based on the first set of digital images and the second set of digital images.
13. 13. The method of claim 12, further comprising: receiving, via the communications network, a third set of digital images and third metadata, wherein the third set of digital images include one or more images of a backside of the creative work, including two or more images of the entire backside of the creative work; and generating, by the one or more processors, the one or more image datasets based on the first set of digital images, the second set of digital images, and the third set of digital images.
14. 14. The method of claim 13, wherein the third set of digital images comprises one or more images of less than the entire backside of the creative work and one or more images of an author signature.
15. 15. The method of claim 11, wherein the comparison dataset further comprises a plurality of feature variable datasets and a plurality of digital image datasets for each additional creative work of the plurality of additional creative works of the comparison dataset.
16. 16. The method of claim 11, further comprising: partitioning, by the one or more processors, into two or more subsets, each of the following: one more feature variables of the creative work, the one or more image datasets, the one or more datasets extracted from an image dataset, and the first set of digital images; determining, using a second machine learning model and based on a comparison of one or more portions of two or more corresponding datasets in each of the different subsets, an external consistency metric of the two or more subsets; determining, using a third machine learning model and based on a comparison of one or more portions of data within the same subset, an internal a consistency metric for a single subset; and determining, based on the output of the second machine learning model and the output of the third machine learning model, a confidence score for one or more datasets associated with the digital images of the creative work.
17. 17. The method of claim 11, further comprising: generating, by the one or more processors, an author profile for a first author of the creative work, the author profile comprising one or more additional creative works of the first author, one or more associated author profiles of one or more authors associated with the first author, one or more time periods, one or more geographic locations, and one or more creative work categories associated with the author; and determining, using the first machine learning model based on the one or more feature variables of the creative work, the author profile of the first author, and the comparison dataset, the authenticity confidence level of the creative work.
18. 18. The method of claim 16, further comprising: determining, by the one or more processors, a creative category associated with the creative work; identifying a one or more associated creative works associated with the creative category; generating, by the one or more processors, an author profde of a second author based on at least a portion of the comparison dataset that corresponds to one or more creative works authored by the second author and a portion of the comparison dataset that corresponds to the one or more associated creative works; and determining, using the first machine learning model and based on at least the one or more characteristics of the creative work, the author profile for the first author, the author profile for the second author, and the comparison dataset, the authenticity confidence level of the creative work.
19. 19. The method of claim 11, wherein one of the one or more feature variables of the creative work comprises a brush sign frequency map for a subject of the creative work.
20. 20. A method, comprising: receiving, via a network, a first set of images of a first creative work; determining, by one or more processors and based on the set of images, one or more first feature variables, each of the first feature variables corresponding to a characteristic of the first creative work; generating, by the one or more processors and based on at least the one or more first feature variables, a first author profile corresponding to an author of the first creative work; obtaining, by the one or more processors, a comparison dataset that comprises one or more author profiles each associated with one or more additional creative works; training, by the one or more processors and with input including the first feature variables and the comparison dataset, a first machine learning model to generate an output indicating an authenticity score of a creative work that corresponds to the first feature variables; and updating, by the one or more processors and based on the output of the first machine learning model indicating an authenticity score of the first creative work, the comparison dataset to include the first author profile.

Description

ARTWORK AUTHENTICITY SYSTEMS AND METHODS TECHNICAL FIELD [0001] The present disclosure relates generally to ascertaining artwork authenticity based on at least digital image data of the artwork and, more specifically, to systems and methods that can generate an indication of authenticity of a creative work using artificial intelligence models and digital image data of a creative work. INTRODUCTION [0002] Some of the most valuable assets in the world are paintings and other creative works. As a result, creative works are often the subject of unauthorized reproductions, elaborate forgeries, or one or more, potentially high-quality, counterfeit copies. Accordingly, before a proposed sale of a valuable creative work, the authenticity of that creative work is very often verified (or evaluated), which may be performed, for example, by an expert in art history and/or artwork authenticity. Experts in creative work authenticity are highly specialized and possess a skillset that requires many years of experience to develop. As a result, the cost associated with an expert analysis can be prohibitively high for a majority of creative works on the market today, which possess substantial values and the corresponding risk of purchasing a forgery. For example, many creative works may have sufficient value to create a need to verify the works’ authenticity but that value may also be insufficient to justify the cost of an expert analysis. For example, for the vast majority of creative works currently on the market, the cost of an expert analysis is more than the entire value of the creative work. Nevertheless, the creative works worth less than the cost of a typical expert analysis may include many creative works worth substantial amounts (e.g., several tens of thousands of dollars). [0003] Expert analysis may be inadequate in some instances, or may otherwise fall short. Experts are inevitably human, and their work (or opinions) can be fallible. For example, as the potential value of a creative work increases, there can be greater tendency (or tension) that human factors come into play, such as relying on motivating factors (e.g., external influence, intemal/extemal motivations) succumbing to prejudice, preference, bias, etc. And there is always a possibility that an expert can simply be wrong. While skill and experience can help to mitigate the human factor, additional methodologies and input for verifying, validating, and/or authenticating are desirable. [0004] Another situation where expert analysis may fall short is when information is lacking, or where time constraints or cost considerations prevent obtaining all information. Consider that the traceability of the location and/or ownership through the years may be unavailable. Lack of information can thwart even the most skilled and/or experienced expert. [0005] Presently available technology to supplement, complement, or supplant expert analysis is presently prohibitively expensive. Sophisticated machines and processes provide additional information as to authenticity, but at a cost that generally is only justifiable for the most expensive of creative works (e.g., greater than USD $250,000, or in some cases greater than USD $500,000). [0006] Accordingly, there is a need for technology to verify an authenticity of many creative works more effectively and efficiently (e.g., at a cost that is lower than an average cost of an expert analysis (appraisal)) and/or that can verify an authenticity of a creative work with sufficient confidence for use in authenticating creative works worth many thousands of dollars. SUMMARY [0007] The present disclosure provides at least a technical solution that is directed to the authentication of creative works (e.g., paintings, drawings, sculptures, etc.), including verification of the author (e.g., contributing artists) of a creative work based on an image dataset that depicts the creative work to be authenticated and a comparison dataset that is based on image data for a plurality of separate creative works. [0008] At least one aspect is directed to a method for verifying an artwork’s authenticity. The method can include receiving, via a network, a first digital image data and a first metadata, wherein the first digital image data captures a first creative work at a first distance and the first metadata indicates an output resolution of the first digital image data and an originality of the first digital image data. The method can include determining, by one or more processors, an expected pixels per inch of the first digital image data, which is based on the output resolution of the first metadata and the first distance. The method can include determining, by the one or more processors, an actual pixels per inch of the digital image data based on a digital image resolution of the first digital image data and a dimension of the first creative work. The method can also include verifying, by the one or more processors, the first distance of