Applications of Deep Learning

Source: Deep Learning on Medium

Deep Learning has taken a crucial role in society in different verticle sector. For complex problems with high dimensionality. Deep learning has emerged at the forefront of nearly every major computational breakthroughs in the last few years. Today the power of deep learning helps computers achieve superhuman capabilities.

In my previous article, I explained about some application of Deep learning -Machine learning. Thus, In this article, I add new applications in deep learning.

1. Creating Fake News

The University of Washington trained a Recurrent Neural Network to make fake videos of former President Obama speaking. This demonstrates the capability of people using deep learning to spread fake news and is something that we need to certainly be aware of in the future.

2. Text detection

The computer vision community has been giving lots of attention to the problem of text detection and recognition in images over the past decade. Text detection can be classified in 3 ways.

  • Document Text
  • Graphic Text
  • Optical character recognition

3. Restore colours in B&W photos and videos

Deep learning can be utilized to colourise black and white images.

Videos contain highly redundant information between frames. Such redundancy has been extensively studied in video compression and encoding, but is less explored for more advanced video processing such as colourising a video.
— Sifei Liu.

B&W convert to colourise

4. Real-time multi-person pose estimation

Deep Learning networks can now incredibly help illustrators in evaluating the postures of people.

5. Changing the gazes of people in photos

In video-conferencing and other visual correspondence situations, the participants watch the presentation instead of the camera. This turns the gaze direction away from the other party and wipes out eye to eye connection. Deep Learning network to change the gaze of the person.

6. Music composition

Voice recognition can likewise be utilized to Deep Learning network to produce music compositions.

7. Transferring style from paintings

Transferring the style from one image onto another can be viewed as an issue of texture transfer. In texture transfer, the objective is to combine a texture from a source image while constraining the texture synthesis in order to preserve the semantic content of a target image.


I think deep learning may be a deep paradigm change in outlook in that we’ll build software that is not perfect nor very efficient however which will learn and theoretically feel what we do. I hope you enjoy reading about this article with the applications of Deep Learning.