Three months ago I joined 16 thousand other students in participating in the Intel® Edge AI Foundation Course on Udacity. Of those 16 thousand, 750 were chosen to receive a scholarship to the full Nanodegree program. I am happy to announce that I was selected! 😁
Foundation Course Review
The foundation course consisted of four main modules that gave an overview of the Intel® Distribution of OpenVINO™ toolkit.
Leveraging Pre-Trained Models
This module discussed the Pre-Trained Models available in the Open Model Zoo of the OpenVINO toolkit. I learned how to load these models using the Model Downloader utility as well as how to pre-process inputs required by the model.
The Model Optimizer
All models need to be converted to and Intermediate Representation (IR) in order to be executed by the tookit. In this way many different frameworks are supported such as Caffe and TensorFlow. The Model Optimizer is a utility for converting models created by these frameworks into an IR. In this module I learned how to convert various models from the supported frameworks to IRs using this tool
The Inference Engine
The Inference Engine is a library of classes that accept input data and execute the model in IR format on devices. I learned how to perform synchronous and asynchronous inference as well as how to handle the results.
Deploying an Edge App
In this final module I combined my previous learnings to set up an application that streamed video to a server using FFmpeg where live inference could be performed on the scene. Results of the inference were then sent to a frontend application via MQTT.
The foundations course provided a comprehensive taste of what the Intel® Distribution of OpenVINO™ toolkit has to offer and I am excited to continue my studies in the rest of the program! 😄