NEW YORK– Computer vision company LexSet, a startup company which uses AI to identify and describe products such as furniture, has recently won several awards for its visual technology.
The company won the pitch competition for Startup to Watch at the Augmented World Expo, as well as the Startup Lab Pitch competition at GS1 Connect, a conference on bar codes, QR codes and all things product identity. Data Con LA, a conference on data science, named LexSet the best Startup Showcase, and most recently the company won the popular vote of the Emerging AI Pioneers Showcase at the O’Reilly AI Conference.
“We built an engine using Hollywood special effects techniques to generate thousands and thousands of images from 3D models of objects,” said Leslie Oliver Karpas, president and co-founder of LexSet. “Working from 3D scans or the models that those objects were manufactured from, we can generate infinite amounts of training data curated to any particular use case.”
LexSet simulates real-world objects and environments from 3D content using its photo-realistic synthetic data engine.
The company’s synthetic data engine has turned more than 80,000 files into 500 million rendered images of furniture and other interior architectural elements in thousands of different angles and in different lighting conditions in order to maximize the system’s ability to recognize those objects when presented with a photo or a real-world space, Karpas said.
LexSet offers the ability to find products based on inspirational images and offers both visual search and automated tagging. The visual search function can be integrated into a website or a mobile app’s search bar to automate the process of making the products shoppable. An additional form of visual search allows a widget to be placed onto a search bar so when customers upload an image; the system will then identify all the existing objects in the image and match them with similar products from the brand’s catalog.