Shoes are the most counterfeited products in the world so if you’re looking to get a pair of Nike Air Force 1 on a second-hand online marketplace such as Craigslist or eBay, how can you trust they are real? Sneaker-selling startup GOAT is fighting the problem by using AI to verify the shoe’s authenticity.
“When you smell a Nike factory shoe, it has that distinct factory smell. And when you smell a fake Nike, it has that fake artificial smell,” Eddy Lu, co-founder of GOAT said. “If you’ve smelled enough sneakers, you know that smell.”
Using deep learning, GOAT helps buyers and sellers verify shoe authenticity with a “ship to verify” model. When a shoe is purchased the seller sends it to GOAT to verify. Once authenticated, the company releases the funds to the seller and ships the product to the buyer.
Using NVIDIA TITAN Xp GPUs and NVIDIA Tesla GPUs on the Amazon Web Services Cloud with the cuDNN-accelerated PyTorch deep learning framework the company trained their neural networks on 75,000 images of authentic sneakers.
The company also collects data points such as color, suppleness of the sole, hardness of certain rubbers, texture, and the quality of the seams, the company said.
“We use a lot of different heuristics and data points to verify whether the shoe is authentic or not. We see the most sneakers in the world. And we are the leader in data collection around sneakers,” said Andy Shin, the company’s Chief Technology Officer. “So every shoe that comes in, we’re collecting all the data points for it. So, we actually, we know what fakes look like. And we know what real shoes look like.”
The company has 400 employees, and 60 of them are engineers and data scientists.