How Deep Learning Allowed Computers To See

We want our smartphones and computers to be as advanced as possible. But what exactly is this level of advanced that we so desire? It is the perfection of our brain we want our devices to have. That is only possible through Artificial Intelligence. No other kind of coding can make a device smart enough to recognize and differentiate the way we do. On similar lines, we also want our devices to see. Now how could such a thing be possible? Through deep learning we have achieved this technological marvel.
To understand the how and the why, we have to understand the concept of machine learning. Machine learning is the process of teaching decision-making to a computer. That is done by teaching it through substantial data to help correlate between subjects. So, in the case of image recognition, we have to feed/load a number of images. An image itself is a collection of a lot of data points-pixels. So, try to comprehend how a device would be able to differentiate and recognize objects based on a pattern of pixels. Deep learning helps in this case. We need a lot of images and images themselves are collection of data points. There are complex neural networks that have to extract features using algorithms. We have taught our computers how to see like a three-year old child at best till date.

Neural networks explained through a Machine learning point of view
Image source:
FeiFei Li, director of Stanford's AI lab and her team, had started the Image Net project by the help of which they collaborated with more than 50,000 people in 180 countries to create the largest image database in the world. This database of images had around 15 million labeled and differentiated images. It explores around 22000 categories. Because of AI and Deep Learning teams working on such databases, computers can now see without human help. The neural networks involved in a giant project have over thousand million nodes. So, computers can train themselves with the help of such neural networks. No, it is not like terminator. They cannot have a human consciousness till taught. Even then there would be several limitations and issues involved. So, with the help of deep learning, data science and other field computers can see. This image recognition has even started being implemented in mobile OSs making use of ML.

Neural networks explained with the help of deep learning
Image source:


Every computer and mobile device today, is making use of Machine or Deep learning in some way or the other. From image recognition, classification, AI has even generated images. The generated images do not make sense to us, but it is the result of neural networks. Because of the black box problem, we cannot see what is going on inside such a sophisticated network. It is like trying to figure out how each twig in a haystack is arranged. Deep Learning Training Kolkata centers can help you out with further explanations. You can also enroll in our Learn Deep Learning Kolkata centers.


Are You A Deep Learning Aspirant?


Leave a Reply

Your email address will not be published. Required fields are marked *