How to Operationalize and Scale Analytic Pipelines with Computer Vision Models at the Edge
Many industries are turning to cameras and computer vision to solve novel and non-traditional challenges that may be hard to accomplish with traditional data sets. Example use-cases include biomedical imaging in Healthcare, smart cities for governments, and automated defect detection in manufacturing. However, just having a computer vision model is not enough to generate the necessary insight to drive action. There is a need to have a complete analytic pipeline that has all the necessary components to transform input data sets into actionable insights. This analytic pipeline must have sufficient accuracy to have confidence in predictions, minimal latency to provide these decisions in a timely manner, and have the ability to deploy and scale to multiple cameras and locations. Come and see how this analytic pipeline for a manufacturing use-case can be built and operationalized using the SAS Platform. We will show how the analytic pipeline consumes multiple data sources, contains multiple post-processing analytic techniques to interpret and generate actionable information from the computer vision model outputs, leveraging technology optimized for compute, and the intermixing of both SAS and Open Source technology in the same analytic pipeline.