Mar 13, 2019
Computer vision is like human vision with superpowers. From autonomous vehicles to new retail applications, it's opening new vistas of opportunity. But to realize them requires rigorous training. Listen as Matthew Quinlan, Principal Director of Geospatial Platforms, Accenture Software & Platforms and Principal Director of Geospatial Platforms, Accenture Software & Platforms and Global Geospatial Platforms Lead at Accenture and Daryn Nakhuda, CEO of Mighty AI, explore the challenges involved in making sure that computer vision algorithms are trained using the data and machine learning techniques they need to operate in complex visual environments. In this podcast, you'll hear about the wide range of factors that need to be considered when it comes to realizing computer vision's potential. Identifying the terabytes of data required for a specific use case is just the start. Synthetic data can augment visual information captured from real world environments. And all that data has to be labelled precisely, and at scale, before it can be loaded into a deep learning framework to start creating models for testing.