AI captures lustre of Ajanta cave paintings in cyberspace

Ajanta_Painting2.jpg

Sapio Analytics, Mumbai and Ajantahc, London, who partnered with IITs and several Indian universities have using tools of Artificial Intelligence (AI) and deep learning, restored some of the damaged murals to their pristine elegance before storing them on their hard drives, for sharing with the global digital community

A group of top digital experts have jumped geographical boundaries and “captured” the glory of Ajanta paintings in cyberspace, for connoisseurs of art across the globe to appreciate and applaud.

Sapio Analytics, Mumbai and Ajantahc, London, who partnered with IITs and several Indian universities went beyond simply digitising and storing some of the heritage paintings. Instead, using tools of Artificial Intelligence (AI) and deep learning, they have restored some of the damaged murals to their pristine elegance before storing them on their heard drives, ready for sharing with the global digital community.

"There are roughly 650 sq. mts (or, 10,00,000 sq. in.) of paintings in 30 caves, with four caves considered ground-breaking in terms of their artistic value at Ajanta," Sapio Analytics's CEO Ashwin Srivastava told Indo Asian News Service (IANS).

Restoring the painting and digitizing them was essential. Since colour paintings adorning the walls and ceiling of the Ajanta Caves date from around 2 BC, or over 2200 years ago, many, over the centuries have been damaged.

Digital restoration of the murals was a painstaking collective effort.  IIT’s, including Roorkee IIT pitched in with its AI expertise to render the paintings their authenticity.

Already, the collective effort to showcase Ajanta’s spectacular art to the rest of the world has paid off. In November 2020, these paintings got a “virtual” entry into the Arctic World Archives (AWA) on Norway's Svalbard Island. Located deep inside a remote, icy Arctic mountain, 970 kms from the North Pole, AWA has been declared a demilitarized zone by 42 countries.

The “Renunciation of King Mahajanaka" - an early 'avatar' of Lord Buddha - in Cave No 1, clicked using zero-light echnology by renowned photographer Benoy K. Behl, starred in the specially curated digital film format at AWA.

Backstage, Srivastava explains how the digital restoration of the murals was conducted. "We select a mutilated area of the mural through image processing-based identification using YOLO Machine Learning Model, restored it by altering the pixels of the damaged elements through mathematical inference of the neighbouring undamaged portions."

High-end pixel-level scanning of the original artwork was carried out. Deep learning algorithms were applied based on the “feedstock” provided by art historians such as Behl.  The hi-tech enterprise is also spinning off unintended takeaways. For instance, an 'academic catalogue of jewelleries’ used thousands of years ago has been compiled using AI technology.

But the team’s herculean efforts are only a start.  Ajantahc's Director B.V. Choolun points out that  only 10 sq. mts. (or, 16,000 sq. in.) have so far been digitally restored for posterity.

Srivastava adds that ceaseless efforts are required as time is running out to save in cyberspace, the artwork in Ajanta caves, rediscovered' 202 years ago in April 1819 by John Smith, a British officer.

(With inputs from IANS)

Covid-19 In India