London, United Kingdom – Artificial intelligence (AI) has proven once again that it can uncover details hidden from the human eye, this time in a Raphael painting. An AI neural network has detected an anomaly in the face of St Joseph in the painting known as the Madonna della Rosa, suggesting that it may not have been painted by Raphael himself.
For years, scholars have debated the authenticity of the painting attributed to Raphael. While determining the provenance of an artwork requires a multifaceted approach, researchers from the UK and US have now introduced a novel method based on AI algorithms to shed light on the mystery.
Employing a custom analysis algorithm, the researchers trained the computer using authenticated Raphael paintings to discern the unique style and characteristics of the renowned Italian master’s work, down to the finest details like brushstrokes, color palettes, and shading. This deep feature analysis delved into a microscopic level that surpassed human vision capabilities.
The team utilized a modified pre-trained architecture developed by Microsoft, known as ResNet50, alongside a traditional machine learning technique known as a Support Vector Machine. This approach had previously demonstrated a 98 percent accuracy rate in identifying Raphael paintings. However, in examining individual faces within the Madonna della Rosa, the AI highlighted a discrepancy – while the Madonna, the Child, and St John bore the hallmarks of Raphael, St Joseph’s face did not.
The face of St Joseph has long been a point of contention in debates over the painting’s authenticity, with some critics suggesting it was not on par with the other figures in the artwork. The research team’s findings support the notion that an unnamed artist, potentially Giulio Romano, a student of Raphael, may have had a hand in creating St Joseph’s face.
The Madonna della Rosa, believed to have been painted between 1518 and 1520, has remained under scrutiny since the mid-1800s when doubts first arose about its attribution to Raphael. While the AI has helped uncover the painting’s secrets, the researchers stress that it is meant to assist art experts rather than replace them in the authentication process.
In the ever-evolving landscape of art analysis, modern technology continues to unveil new insights into classic paintings, showcasing the potential for AI to augment traditional methods of art authentication. This study, published in Heritage Science, underscores the symbiotic relationship between AI technology and human expertise in unraveling the mysteries of art’s past.