Eugene Mezhibovsky, Senior Systems Architect at Coherent Logix, presents the “Using hxGraph and OpenCV to Implement Industrial Inspection Applications on the HyperX Processor” tutorial at the May 2021 Embedded Vision Summit.
hxGraph is a high-level tool that enables users to develop computer vision applications using the OpenCV APIs and then target the application to run on the high-performance HyperX embedded processor. hxGraph accepts a Python program written using the OpenCV API syntax and generates code to implement the graph (pipeline) on the HyperX processor using a library of OpenCV functions.
In this tutorial, Mezhibovsky demonstrates the software framework using an industrial inspection application. He shows the construction of the processing pipeline in OpenCV using Python, initially running the application on the host computer. He then targets the HyperX processor and executes the same processing without having to write any new code. Mezhibovsky then demonstrates both processing computer-hosted images and streaming GigE Vision camera images directly to HyperX. He also shows performance benchmarks such as latency and throughput for the application.