VIDEO
HIGHLIGHT
Log InSign up
Can AI detect forged art? | Carina Popovici | TEDxNuremberg
SummaryTranscriptChat

Can AI detect forged art? | Carina Popovici | TEDxNuremberg

Introduction

The speaker introduces a painting and discusses the process of authenticating works of art.

Authenticating Works of Art

  • Authentication is slow, expensive, and subjective.
  • Experts require seeing the original artwork for authentication.
  • Different experts can have contradictory opinions.

Using AI to Authenticate Art

The speaker discusses how artificial intelligence (AI) can be used to authenticate works of art.

Training AI to Recognize an Artist's Style

  • AI can learn an artist's unique style by analyzing photographs of their authentic works of art.
  • Ideally, at least 200 images are needed to create a reliable data set.
  • The AI learns not only brush strokes but also other elements such as color use and composition.

Comparing Features for Authentication

  • Once training is completed, the AI can compare features learned during training with those identified on new works of art.
  • Based on this comparison, the AI returns a probability for the new artwork.

Advantages of AI in Art Authentication

This section discusses the advantages of using AI for art authentication.

Benefits of AI in Art Authentication

  • AI is fully unbiased and has no emotions, making it less subjective than human experts.
  • The training process can take several days, but checking whether a new art piece is by the same artist is just a matter of minutes.
  • After waiting for around three to four days, you will receive an answer to your AI authentication query.

Authenticating Art with AI

This section explains how to authenticate art using AI.

Submitting an Image for Authentication

  • To submit an image for authentication, the program needs about three days to learn the main characteristics from around 700 training images.
  • Once the program has learned these characteristics, it takes less than five minutes to calculate the probability for the authenticity of a new art piece.

Results and Heat Map Visualization

  • The program evaluated the painting as authentic with a 97% probability score.
  • A heat map is used to visualize results on the painting itself. Regions highlighted in bright red have been identified during training as characteristic for the style of the artist, confirming authenticity.

Conclusion

This section concludes that technology offers huge possibilities and every museum can now verify their collections in a fast, affordable and objective manner.

Preserving Cultural Heritage

  • Every curator can find out immediately if works they are bidding on are real or not. This will help preserve our cultural heritage.
Video description

When we go to a museum, we see beautiful paintings displayed in its halls, and we naturally assume that they are authentic. While admiring the paintings, the viewer may never realize how much effort the museum has put into verifying their authenticity. Carina shows how this process can be simplified through the use of modern technology. The audience is invited to a presentation about using Artificial Intelligence to authenticate works of art, and its relevance to the preservation of our cultural heritage. - Video produced by froehlichfilm.de - Carina Popovici is a co-founder and CEO of Art Recognition, a Zurich based ArtTech startup which offers to the art world an AI Engine for art authentication and forgery detection. While being an art lover, she is also an experienced programmer, having developed in the past AI algorithms with applications in physics and finance. Carina received a PhD in Theoretical Physics from the University of Tübingen, and has continued her career as a Research Fellow in Physics. She then moved on to work as a quantitative risk specialist in the Swiss banking industry, before founding Art Recognition. Now, Art Recognition is the only company worldwide offering an AI authentication service commercially, and one of the few startups operating at the forefront of AI, entirely created and led by women. This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at https://www.ted.com/tedx