Technical Content Writer at AlmaBetter at almaBetter
We have heard a lot about computer vision but before getting into coding, it’s very important to master the fundamentals in a layperson’s terms. This blog marks the beginning of a series of articles covering the basics of Computer Vision.
Let’s start by building intuition.
If you were told to name certain things that you’d find in a mall, you’d casually mention things like ice cream stands, clothes, fun-zone, etc. It’s an easy task for humans. However, inside our brains, a complex and dense process takes place that allows us to perceive the things we see.
Our human vision does abstract thinking to understand concepts through trillions of interconnected interactions with the outside world. And similarly, computers also have the concept of vision which is built from complex reasoning and code sources.
Let’s understand the similarities between how a computer and human vision process their own thinking, respectively.
Firstly, let’s start with a question - How will you identify oranges from a basket full of different kinds of fruit?
You will probably say something like: “I see oranges with different shapes, sizes, texture, and pattern. It doesn’t even matter if that orange is painted or it is a toy, you will still call it an orange.”
Well, that’s true. If we consider this case and understand how we assume biologically, we are just classifying the object based on our past experiences. We have stored information over time about how a particular object looks while focusing on texture instead of focusing on colour as we can find oranges of different colours but the shape, size, texture, and pattern remains the same.
What if you were asked to recognize a fruit that you have never seen before?
Chances are you will not be able to name it.
Here, the process of classifying or identifying plays an important role in determining an object. With respect to humans, the perspective is that we have stored information about oranges in our brains on the basis of their shape, size, texture, and pattern.
If you have observed, we don’t focus on colour as we can’t say that an apple is always red because apples could have different colours, so we focus on the other specifications to recognize what kind of object it is.
Computer vision also works with a similar type of understanding when determining an object.
We want to train a system to identify oranges from a fruit basket.
For that, we have to consider all kinds of oranges, their size, shapes, and colours. Even after that, chances are we might have missed out on some important aspects that will fail the system to recognize the case.
Suppose, we present to you a green orange and ask to identify the fruit. It’s guaranteed that the answer is “Orange”.
Reason: The information about an orange’s shape, size, texture, and pattern is instilled in your brain early on. No matter what colour the orange is, you will identify it as orange.
In the case of computer vision, it is about the way you represent the object. If you only train it to recognize oranges, it will fail against identifying apples, since it has never seen it before.
So, what do we know from the above example?
Learning is not about remembering scripts and concepts. It is about grabbing insights and preparing solutions to decode problems.
Just with understanding of oranges, we are quick to decide if the fruit is an orange or not.
Similarly, with the right set of rules and data, computer vision can recognize patterns on par with a human brain.
If you are interested in exploring the world of Data Science, sign up for AlmaBetter’s Full Stack Data Science program. New batch starts soon!
Read our recent blog on “Data Preprocessing with Scikit-Learn: A Tutorial”.