Abstract: 

Welcome to the world of augmented reality and artificial intelligence! In this workshop, we demonstrate AR and AI via hands-on exercises where you will interact with your augmented world. We describe some of the novel applications of AR+AI, their limitations, and social impacts amplified by the COVID-19 pandemic. We leave attendees armed with code, inspiration and ethical considerations for future projects.
In this workshop, you will build an application where: you will take your own snapshot using your webcam and use deep learning to augment your background; you will use deep learning again to perform pose estimation in your augmented world. We will discuss ethical challenges associated with using deep learning and provide resources and guidance on this topic.
In deep learning, we will focus on two techniques: keypoint estimation using a deep neural network, and segmentation. Keypoint estimation recognizes specific points in an image – for example, the locations of joints on the body. Image segmentation separates parts of an image, pixel-by-pixel. For example, segmentation can be used to identify the background pixels in an image.
We will set you up for success to take on challenging projects in this domain. A clear and precise instruction sheet will guide you through exercises ranging from beginner to intermediate level. This workshop is highly interactive with hands-on exercises in which you will incrementally build the pieces of an AR+AI application and test it out. The presentation that accompanies the exercises demonstrates how powerful applications ranging from entertainment to healthcare can be built using the same components that we taught.

Bio: 

Sarah Mohamed is a Software Engineer at MathWorks working on deep learning interoperability. She develops tools that allow AI models to be exchanged and deployed among frameworks such as MATLAB, Pytorch, TensorFlow, and ONNX. In the past, she has also worked as an Application Support Engineer, where she interacted directly with customers developing deep learning applications. She obtained her M.Sc. in Computational Science and Engineering from Harvard University, and holds dual bachelor's degrees in Computer Science and Biology from the University of Virginia. She has experience organizing and delivering hands-on workshops at the Grace Hopper Celebration, the world’s largest conference for women in computing, as well as TechTogether Boston, the largest student-led hackathon for women and nonbinary individuals in Boston.