Making a Robot Eye with JS or Magic of Computer Vision Unraveled

Have you ever wondered how computers see images? Do you want to work on an AR (Augmented Reality) project or build a NodeBot that can recognize and track your face? How about making your own Instagram-like app or finding similer images in Pinterest collection? When you work with an image as your data input, sooner or later you'll encounter the magic term Computer Vision. It can be intimidating to get started. At first, that led me to choose a common and powerful library that works like magic in an unfamiliar language (C++). It felt like the library was a black box and I had no control. I started to wonder, what is so magical about computer vision? After all it is just code someone wrote. Can we unravel it? and reconstruct in JavaScript?? The answer is YES. In this talk, we'll start by getting image data from canvas object, then transform the data to alter an appearance of the image. You will get insight into what data looks like to a browser (hint: it's just an array of numbers !). Next we’ll talk about how to interpret and analyze the data in a meaningful way. Finally, we'll apply what we learned and go through simple image analysis process to build your own AR app in vanilla JavaScript!