This course provides an overview of Computer Vision (CV), Machine Learning (ML) with Amazon Web Services (AWS), and how to build and train a CV model using the Apache MXNet and GluonCV toolkit. The course discusses artificial neural networks and other deep learning concepts, then walks through how to combine neural network building blocks into complete computer vision models and train them efficiently.
This course covers AWS services and frameworks including Amazon Rekognition, Amazon SageMaker, Amazon SageMaker GroundTruth, and Amazon SageMaker Neo, AWS Deep Learning AMIs via Amazon EC2, AWS Deep Learning Containers, and Apache MXNet on AWS. The course is comprised of video lectures, hands-on exercise guides, demonstrations, and quizzes.
Each week will focus on different aspects of computer vision with GluonCV. In week one, we will present some basic concepts in computer vision, discuss what tasks can be solved with GluonCV and go over the benefits of Apache MXNet.
In the second week, we will focus on the AWS services most appropriate to your task. We will use services such as Amazon Rekognition and Amazon SageMaker. We’ll review the differences between AWS Deep Learning AMIs and Deep Learning containers. Finally, there are demonstrations on how to set up each of the services covered in this module.
Week three will focus on setting up GluonCV and MXNet. We will look at using pre-trained models for classification, detection and segmentation.
During week four and five, we will go over the fundamentals of Gluon, the easy-to-use high-level API for MXNet: understanding when to use different Gluon blocks, how to combine those blocks into complete models, constructing datasets, and writing a complete training loop.
In the final week, there will be a final project where you will apply everything you’ve learned in the course so far: select the appropriate pre-trained GluonCV model, apply that model to your dataset and visualize the output of your GluonCV model.