The COVID-19 pandemic has thrust ecommerce to the forefront of retail. It has also made inroads in the fashion industry, where online orders have never been the main sales channel. Many consumers who have not used online channels to shop for clothes or shoes are coming round to this form of shopping. However, while online shopping becomes a new reality in the fashion industry, retailers are facing a wave of over-returns, disrupting replenishment and their supply chains.
Shoppers don’t know if an item they’re about to purchase online will suit them until they try it on at home. Will the colors match their complexion? Will this jacket go with those slacks? Will they look like the millions bucks they imagine? If not, they may return the item or forego the transaction altogether, contenting themselves instead with merely browsing.
In response to this challenge, deepsense.ai has developed an AI-powered fitting room, where customers can see an online simulation of their appearance in a new piece of clothing. The system also can suggest appropriate sizes.
Our multi-step deep learning-based system detects and transfers the clothes from model to online client simulation. The system analyzes two images, the source - the clothes to be tried on, and the target - the shopper’s image. We use key-point detection, instance segmentation and image warping to produce an initial impression, which is then further polished by a specialized Generative Adversarial Neural Network (GAN).
The virtual fitting room solution may increase online sales by up to 11% while reducing the rate of returns by 19%.