Getting accurately labeled training data remains a challenge for computer vision and AI startups competing against the world’s Googles, Facebooks, Apples, and Microsofts, all of which have access to exponentially growing boatloads of the stuff. So why go for the real thing when an artificially created version might also do the trick? Advances in synthetic data which runs the gamut from photorealistic images to VR and video game simulations are such that the “simulation-to-reality gap is fast disappearing,” according to Cornell Tech professor Serge Belongie, who is one of many experts and companies that will discuss synthetic data, among other visual AI topics, at next week’s LDV Vision Summit, which focuses on the latest innovations, trends and startups in the computer vision space. Go, and get some real news about fake data.
sign up for our newsletter