5 Ways You Can Learn Computer Vision

Source: Artificial Intelligence on Medium

2. YouTube Videos

I realize not everyone has the time or funds to pursue a master’s degree, but the next best thing, and probably the cheaper option is to utilize the internet.

YouTube is a great learning platform, and I watch ML/CV content from popular channels to further my understanding and sharpen my practical skills.

Here are a few examples of YouTube channels that help me learn Computer Vision.

This channel content is second to none, and I can’t praise this channel’s creator enough.

Those who are familiar with 3Blue1Brown will agree with me when I say that he manages to teach topics that would take University lecturers hours of teaching in 15–20 minutes short videos.

The channel taught me the basics of Machine Learning and Neural networks and made maths simple.

Within machine learning, it is essential to understand topics such as linear algebra, calculus, and partial differentiation. When studying Neural Networks, it is crucial to understand the fundamental components of a neural network.

It is also essential to understand concepts such as backpropagation, vanishing gradients, and different neural network architectures.

The playlist below from 3Blue1Brown is a must-watch if you would like a simple explanation of the processes that occur within neural networks.

After understanding the basics of ML and Neural networks, you can dive into some academic sessions from Stanford that explains some common Computer vision techniques and algorithms.

This playlist provides you with the step by step academic knowledge of techniques that enable image classification, object detection, reinforcement learning, and plenty more.

Completing this playlist will provide you with a piece of general knowledge on Computer Vision algorithms, techniques, and approaches that are currently in use today in systems such as self-driving cars, facial recognition applications, etc.

Be sure to take notes and remember the key terms and concepts presented.

If you have a solid grasp of the theory and maths behind neural networks and some machine learning algorithms, you can move on to some practical projects and coding with a common machine learning library (PyTorch) and programming language Python.

PyTorch is a machine learning framework that provides people like you and me with the tools to implement and launch machine learning solutions and neural network architectures from the comfort of our laptops. PyTorch is most researcher’s framework of choice when it comes to implementing machine learning solutions.

Python is a modern programming language that has a place in various computing-related disciplines such as web development, server development, software development, etc. The language is prevalent and can is utilized for the implementation of machine learning solutions.

Deep Lizard’s video playlist will gradually take you through the process of neural network implementation. You will learn about topics such as the utilization of GPUs for neural network training, data structures (array and tensors); neural network layers; hyperparameters, and much more.

With all the knowledge you have accumulated, you might be able to start a ML project, and after that, have the basic knowledge to get an entry-level computer vision job position.