We are excited to share a lecture that Nathan Cavaglione, an algorithm engineer at Datagen, gave at IMVC 2020. IMVC is a conference dedicated to the latest in machine vision. It is a prestigious platform for industry and academic leaders in the field and consistently highlights groundbreaking work by global experts. We are proud to have contributed to this year’s conference.
In his lecture, Nathan explains some key concepts and ideas about Simulated Data. With the help of two case-studies focused on object detection and 3D pose estimation, Nathan shows how Simulated Data is created and used in different contexts. The results of these case studies show the promise of Simulated Data to effectively train neural networks, in some cases, more effectively than Manual Data. At Datagen, we are working to generalize these case studies and build entire libraries of visual assets that we believe can help solve the data bottleneck.