Operationalizing Computer Vision at the Edge: The role of a complete analytics pipeline
Session Abstract:
The key aspects of computer vision include methods for acquiring and processing videos and images, training deep learning models, and operationalization of these models. Deep learning architectures available for model training have always been at the heart of any computer vision related discussion. However, as more organizations start to adopt computer vision as a strategy for additional business insights, three related questions have come to the forefront – First, what if organizations don’t have access to enough data that allows for testing a computer vision model before deployment? Second, how are computer vision models deployed, managed, and scaled at the edge? And third, how do computer vision models interplay with broader analytics use cases in organizations? In this presentation, we will address these questions by focusing on the emergence and benefits of synthetic image data, the necessity of a repeatable framework to develop and deploy next-gen models, and the role that a broader analytics framework plays in the context of computer vision.
Speaker:
Juthika Khargharia, Ph.D is a Principal Solutions Architect working for SAS’ Internet of Things Division specializing in machine learning, deep learning and artificial intelligence. In her role, she assists customers in defining their business problems and uses SAS advanced analytics solutions to help them reach their business goals and objectives. Juthika holds a Ph.D. in Astrophysics from the University of Colorado. Her Ph.D. research work consisted of designing observations, collecting and analyzing low signal-to-noise image data obtained from ground based telescopes to determine black hole masses in our Galaxy.