Art Recognition

Art Recognition was established in 2019 by Dr. Carina Popovici and Christiane Hoppe-Oehl. The foundation of the company was driven by the long-standing challenge in the art world of authenticating paintings, a process traditionally reliant on expert judgment, historical research, and scientific analysis. Recognizing the limitations of existing methods, the co-founders were motivated by technological advancements in digital imaging and pattern recognition algorithms in the field of art.

These technological advancements, particularly in the realm of high-resolution digital imagery, enable a more detailed examination of artworks.{{Cite news |title=New tools are making it easier to authenticate paintings |url=https://www.economist.com/books-and-arts/2021/01/07/new-tools-are-making-it-easier-to-authenticate-paintings |access-date=2024-02-10 |newspaper=The Economist |issn=0013-0613}} By analyzing brushstrokes, signature patterns, and other distinct characteristics, and comparing them with known works by the same artist, digital tools offer a new dimension in authentication. Popovici and Hoppe-Oehl aimed to develop an advanced algorithm that could further assist experts by identifying stylistic elements and patterns unique to individual artists, thus aiding in the art authentication process.

File:REPORT example Cezanne Boy in a red vest ENGLISH. ENCRYPTED.pdf

Art Recognition employs a combination of machine learning techniques, computer vision algorithms, and deep neural networks to assess the authenticity of artworks.{{Citation |last1=Schaerf |first1=Ludovica |title=Art Authentication with Vision Transformers |date=2023-07-10 |arxiv=2307.03039 |last2=Popovici |first2=Carina |last3=Postma |first3=Eric}} The AI algorithm analyzes various visual characteristics, such as brushstrokes, color palette, texture, and composition, to identify patterns and similarities with known authentic artworks.

The company's technology undergoes a process of data collection, dataset preparation, and training. In the initial phase, datasets are compiled, and data selection is supervised by art historians to ensure the inclusion of genuine artworks by specific artists. This approach aims to avoid including artworks that may have been partially completed by apprentices or contain mixed authorship.

Upon the preparation of datasets, a segment of the image set is used for training the AI algorithm, while the remaining images are set aside for testing. This phase aims to ensure the algorithm's proficiency in distinguishing authentic artworks from forgeries. Post-training, the algorithm undergoes evaluation with the test data, assessing its accuracy and efficacy in authenticating artworks.

After the testing phase, the AI algorithm is applied to analyze new images, including submissions from clients. Additionally, the algorithm is designed to identify artworks generated by generative AI, mimicking the style of renowned artists. This capability equips the algorithm to withstand adversarial attacks, enhancing its reliability in differentiating between authentic and artificially generated fake art pieces.{{Cite web |title=2312.14998 - Synthetic images aid the recognition of human-made art forgeries |url=https://www.emergentmind.com/papers/2312.14998 |access-date=2024-02-10 |website=www.emergentmind.com}}

Art Recognition's collaboration with Tilburg University in the Netherlands has resulted in the acquisition of a research grant from Eurostars, Eureka (organisation) the Eureka's flagship small and medium-sized enterprises (SME) funding program. In addition, the company has formed a partnership with the University of Liverpool in the United Kingdom, which has been supported by the Science and Technology Facilities Council (STFC) Impact Acceleration Award. Furthermore, Art Recognition has established a relationship with Innosuisse, a Swiss innovation agency,{{Cite web |title=Innosuisse Discover 2021 - Recognising art forgeries from a photo |url=https://2021.discover-innosuisse.ch/en/projects/art-recognition |access-date=2024-02-10 |website=InnoSuisse |language=en}} to expand its research and development initiatives.

Art Recognition has formed a strategic collaboration with [https://www.abk-stuttgart.de/person/nils-buettner/ Nils Büttner], an art historian and professor at the State Academy of Fine Arts Stuttgart (ABK Stuttgart). By fostering dialogue between academic researchers and market professionals, the collaboration aims to refine existing authentication practices and introduce scientifically robust methodologies into the art sector.https://www.kunstgeschichte-ejournal.net/622/

In May 2024, Art Recognition played a key role in identifying counterfeit artworks, including alleged Monets and Renoirs, being sold on eBay. The findings contributed to a broader discussion on the role of AI in preventing art fraud, particularly in online marketplaces where traditional expertise is often lacking. The case underscored the increasing importance of AI as a fraud detection tool.[https://www.theguardian.com/artanddesign/article/2024/may/08/fake-monet-and-renoir-on-ebay-among-counterfeits-identified-using-ai]

Germann Auction made history in November 2024 by becoming the first auction house to successfully conduct a sale of artwork authenticated entirely by artificial intelligence. This milestone reflects the increasing reliance on technological solutions in the art market, where AI-driven authentication processes provide data-backed evaluations of artwork authenticity.https://www.artnews.com/art-news/news/swiss-auction-house-is-the-first-in-the-world-to-sell-artwork-authenticated-solely-by-ai-1234725695/

The NHK WORLD-JAPAN news report examined the vulnerabilities of the Japanese art market to forgery, particularly in light of the Beltracchi scandal of the 1990s. During this period, Wolfgang Beltracchi successfully sold numerous counterfeit paintings to Japanese collectors, who only later discovered the deception. This case highlighted the challenges of authentication in the Asian art market, exposing weaknesses in provenance verification and expert assessment. The report explores how advancements in AI-based authentication, including the work of Art Recognition, are now being used to prevent similar fraud. Artificial intelligence is being integrated as a supplementary tool to traditional connoisseurship, aiming to enhance security in the global art trade.[https://www.youtube.com/watch?v=6ks5-YT37GE]

As of January 2025, Art Recognition has appointed art crime expert and Pulitzer Prize finalist Noah Charney as an advisor. Charney, the founder of the Association for Research into Crimes against Art (ARCA), is a leading authority on art forgery, provenance studies, and cultural heritage crimes.

Art Recognition's AI algorithm has received attention from various media outlets and industry events. The company was featured on the front page of The Wall Street Journal{{Cite web |title=Is This a Real Raphael Painting? AI Says Yes, But Humans Aren't So Sure. |url=https://www.wsj.com/story/is-this-a-real-raphael-painting-ai-says-yes-but-humans-arent-so-sure-ede9ed54 |access-date=2024-02-10 |website=WSJ}} for its involvement in the authentication case of the Flaget Madonna, believed to have been partly painted by Raphael.

A broadcast by the Swiss public television SRF showcased how the algorithm can be used to detect art forgeries with high accuracy.{{Cite web |date=2020-10-23 |title=Die Idee - Mit einem Algorithmus Kunstfälschungen erkennen |url=https://www.srf.ch/sendungen/10vor10/die-idee/die-idee-mit-einem-algorithmus-kunstfaelschungen-erkennen |access-date=2024-02-10 |website=Schweizer Radio und Fernsehen (SRF) |language=de}} Additionally, the company's work was featured in a TEDx talk discussing the use of AI in art authentication.

The technology developed by Art Recognition has been recognized for its role in providing a technology-based art authentication solution, compared to traditional methods. This advancement is seen as significant in the field of art verification, offering a modern approach to a historically complex process.{{Cite news |last=Müller |first=André |date=2020-01-19 |title=Art Recognition: Carina Popovici legt Kunstfälschern das Handwerk |url=https://www.nzz.ch/zuerich/art-recognition-carina-popovici-legt-kunstfaelschern-das-handwerk-ld.1534429 |access-date=2024-02-10 |work=Neue Zürcher Zeitung |language=de-CH |issn=0376-6829}}

The use of AI in art authentication, as pioneered by Art Recognition, has become a topic of professional discourse. Notably, this subject was the focus of a debate on Radio Télévision Suisse, where experts deliberated over the capabilities and limitations of AI in identifying art forgeries. Such discussions highlight the evolving landscape of art authentication in the age of digital technology.{{Cite web |date=2021-12-19 |title=L'intelligence artificielle peut-elle détecter les faux dans l'art? |url=https://www.rts.ch/play/tv/forum/video/lintelligence-artificielle-peut-elle-detecter-les-faux-dans-lart-debat-entre-christiane-hoppe-oehl-et-anne-laure-band-video?urn=urn:rts:video:12731224 |access-date=2023-06-23 |website=rts.ch |language=fr}}

Despite the advancements in AI-driven art authentication, the field continues to face unique challenges, particularly regarding the acceptance of such technologies. Experts in the field stress the necessity of using AI as a complementary tool alongside traditional methods, rather than as a stand-alone or definitive solution for authenticating art.{{Cite web |date=2023-03-01 |title=AI Companies Are Authenticating Old Master Paintings, But the Art World is Skeptical |url=https://observer.com/2023/03/ai-companies-are-authenticating-old-master-paintings-but-the-art-world-is-skeptical/ |access-date=2024-02-10 |website=Observer |language=en-US}}

Art Recognition's AI algorithm has been applied to several high-profile and controversial artworks, sparking significant interest and debate in the art world.

1. Samson and Delilah at the National Gallery in London: The National Gallery's "Samson and Delilah", traditionally attributed to the artist Rubens, has also been examined using Art Recognition's AI, which has assessed the painting as non-authentic. This analysis contributed to ongoing scholarly discussions regarding the work's authenticity.{{Cite news |last=Alberge |first=Dalya |date=2021-09-26 |title=Was famed Samson and Delilah really painted by Rubens? No, says AI |url=https://www.theguardian.com/artanddesign/2021/sep/26/was-famed-samson-and-delilah-really-painted-by-rubens-no-says-ai |access-date=2024-02-10 |work=The Observer |language=en-GB |issn=0029-7712}}
2. De Brecy Tondo Madonna. A research team from Bradford University and Nottingham University initially attributed the painting to Raphael, employing an AI face recognition software,{{Cite news |last1=Khomami |first1=Nadia |date=2023-07-14 |title=Painting 'undoubtedly' by Raphael to go on display in Bradford |url=https://www.theguardian.com/artanddesign/2023/jul/14/painting-undoubtedly-raphael-go-on-display-bradford |access-date=2024-02-10 |work=The Guardian |language=en-GB |issn=0261-3077}} while the AI developed at Art Recognition returned a negative result.{{Cite web |last=Harris |first=Gareth |date=2023-10-24 |title=A question of attribution: just how useful can AI tools be? |url=https://www.theartnewspaper.com/2023/09/11/a-question-of-attribution-how-useful-can-ai-tools-be |website=The Art Newspaper}} As face recognition methods have proven to be less appropriate for art authentication, {{Cite web |date=2023-10-23 |title=Two A.I. Models Produced Different Results When Authenticating a Raphael Painting. Here's Why That Doesn't Undermine the Tool's Potential |url=https://news.artnet.com/art-world-archives/ai-authentication-raphael-painting-op-ed-2373793 |website=Artnet News}} the Bradford group developed an alternative AI-based approach similar to that used by Art Recognition. A key distinction between the two systems lies in their training datasets: the Bradford group's AI was trained on 49 images, whereas Art Recognition employed a larger dataset of over 100 images. This difference highlights the role of dataset size and composition in the effectiveness of AI-driven art analysis.
3. Lucian Freud Painting Controversy: Featured in The New Yorker, a painting attributed to Lucian Freud became a subject of dispute. Art Recognition's AI analysis played a pivotal role in examining the painting's authenticity, contributing to the broader discussion about the challenges in verifying modern artworks.{{Cite magazine |last=Knight |first=Sam |date=2022-09-19 |title=The Case of the Disputed Lucian Freud |url=https://www.newyorker.com/magazine/2022/09/26/the-case-of-the-disputed-lucian-freud |access-date=2024-02-10 |magazine=The New Yorker |language=en-US |issn=0028-792X}}
4. Titian at Kunsthaus Zürich: A painting attributed to Titian, housed at Kunsthaus Zürich, has been a topic of debate among art experts. The application of Art Recognition's technology offered a new perspective, utilizing AI to analyze the painting's stylistic elements in comparison with authenticated works of Titian. Following this debate, Kunsthaus Zürich has announced plans to initiate a comprehensive project aimed at resolving the authenticity questions surrounding the painting. This project is set to involve collaboration with scientists and technology companies, leveraging a multidisciplinary approach to authenticate the artwork.{{Cite web |last=Dafoe |first=Taylor |date=2023-03-28 |title=A Zurich Museum Found Out It May Have Acquired a Fake Titian. So Why Did It Buy Another Painting That Looks Just Like It? |url=https://news.artnet.com/art-world-archives/kunsthaus-zurich-titian-authenticity-poussin-2276241 |access-date=2024-02-10 |website=Artnet News |language=en-US}}
5. Art Recognition has contributed to the authentication debate surrounding The Polish Rider, a painting traditionally attributed to Rembrandt but subject to scholarly debate. Utilizing AI-driven analysis, the study examined stylistic features, brushstroke patterns, and compositional details to assess the painting’s authenticity. The findings provided quantitative insights that supported its attribution, reinforcing the potential of AI in resolving long-standing art historical disputes.https://www.theartnewspaper.com/2024/12/01/can-ai-shed-new-light-on-how-much-of-the-polish-rider-was-painted-by-rembrandt

In each of these instances, Art Recognition's involvement has provided additional perspectives through AI analysis while contributing to broader conversations about the role of technology in art authentication. These cases demonstrate the evolving nature of art verification, where traditional methods are being supplemented, and sometimes challenged, by new technological approaches. However, they also underline the ongoing debates about the acceptance of AI in the field of art history, especially in the authentication of works by renowned artists.

{{Reflist}}

• {{Official website|https://art-recognition.com/our-technology/}}
• [https://www.art-magazin.de/ Art magazine website]

{{DEFAULTSORT:Art Recognition}}

Category:Companies based in the canton of Zürich

Category:Swiss companies established in 2019

Category:Art forgery

Category:Artificial intelligence companies

Category:Authentication

Category:21st-century introductions

Category:Data science

Category:2019 establishments in Switzerland