AI-designed clothes sell faster than conventional

Clothes designed using artificial intelligence were selling faster than those designed by humans.

Mr Jayendran Venugopal, chief technical officer of Myntra, said, “In fashion there are long lead times and uncertain customer trends. From a fashion designer concept to the retail sales it would take 14 months. Using machine learning, AI is extracting attributes based social media trends, red carpet and street imagery.” Using tools like generative adversarial network and Rapid, AIdesigned clothes were selling three times faster and fetching two times more revenue, Mr Jayendran told the fifth edition of Nasscom Product Conclave here on Friday.

He said this was being tested for western wear, and would be slowly rolled out to ethnic wear. “There are lot prints and designs when it comes to Indian styles. We are working on it,” Mr Venugopal said. He said the AI technique was being sought by international brands.

The shoe size finder on Myntra is made using Augment reality. Speaking on deep learning, he said, “Machine learning is barebone statistical algorithms, while deep learning is cutting edge. It is not just about recognising objects but being subjective.”

He said deep reinforcement learning would add objective to function. “We have not reached the level of sophistication of the human brain, but we will be there soon. Neuromorphic computing will help draw inference from human brain.”