Trying Out Computer Vision

I’ve wanted to start working with computer vision for the last few years, but for one reason or another, didn’t pursue it. Now I am.

Why computer vision? Because the ability to turn pixels into something meaningful is a major stride towards smarter software. Identifying objects visually has a wide range of useful applications.

I’ve seen countless videos on YouTube about using OpenCV for motion tracking, face detection, and other awesome stuff. The code didn’t seem very verbose or difficult to understand.

So I started by installing OpenCV. Although I don’t like polluting my host system with installed libraries/dependencies and would rather use Docker to handle compilations, I didn’t want to add complexity to something I had never done before, so I just followed the introductory instructions to install OpenCV on Linux.

After building the first demo project successfully, I felt that the first hurdle was over. I could now start building my own computer vision program!

The first step was to use some sort of image for input. It could be loaded from an image or video file (which is great for reproducing results in tests), but also kinda boring, so I used my webcam stream. How hard could it be? Not hard at all.

Ludicrously simple.

Then I checked out some documentation to see how I could apply filters to the image.

End result:

Let’s see what I can cook up after further experimentation!

Leave a Reply

Be the First to Comment!