Motivation

Colors play a vital role in our lives. You may have noticed that a light color scheme can often have a soothing effect on us. Colors can change our thinking and cause action/reactions.

We choose the problem of Image Colorization as it is one of the deterministic problems of Image to Image translation which has been explored in the recent past. Despite there being no unique solution, we have a baseline with which we can compare our model's learning ability.

Due to the availability of different techniques available for Image to Image translation, we can study the parameters which impact the network's ability to recognize shapes and learn different coloring schemes.

The prime importance of this problem lies in the restoration of the historical images and thereby connecting the present with the past. Moreover, image colorization can be used in color restoration tasks as well as enhancement of surveillance feeds.

Moreover, in certain cases, we can colorize the inputs to a Learning model so as to enhance its ability to learn different features more accurately.