Mindtech Global, developer of a platform for the creation of synthetic data for training AI, has published details on how machine learning engineers can radically speed up the time it takes to train their AI vision systems to see and understand the world around them.
Synthetic images generated on a platform like Mindtech’s Chameleon are produced ready-annotated. If working solely with real-world images, a machine learning engineer can spend up to 80% of their time making sure they are labelled correctly so their AI does not get confused. Automating this task reduces their workload substantially—and alleviates several other issues too, according to Chris Longstaff, VP Product Management, Mindtech Global.
“We encourage visual AI developers to strive to vastly reduce the number of real-world training images they use – by utilising many more synthetic, computer-generated images to train their machine learning networks.
“The reason? In a synthetic imaging platform like Mindtech’s Chameleon, we’ve created the 3D virtual world, we know exactly what objects are in it, and we know where they are in three dimensions – so the image can be annotated automatically and in an instant. On top of that, the synthetic images are, of course, privacy compliant, too.”