The problem ENHANCINATOR solves

We all live in the 21st Century, a place, a time where we all are completely dependent on the internet but we all face one most common problem, that is internet connectivity and availability. People in the rural areas and well as many people in cities and towns have limited data, face an issue of low bandwidth and poor connection, and hence are not able to reap the benefits of the internet. In today's world videos is a major source of information be it in the domain of education, news, sports, and everything thing that one can think of. Comprehending a video in low quality is a really tough job and destroys the user's video streaming experience.

-Unavailability or bad network quality in many places.

-Since online video streaming is a major source of entertainment and information, watching videos in bad quality is often useless as most of the details will not be visible, hence it becomes a necessity to stream videos in good quality.

-In today's world, where COVID-19 has taken over the entire globe, education, work, and all the other activities are taking place online through live video streams. There are often network issues that hamper the work. This obviously should not be an obstacle in one's path to success.

Solution

In order to solve the above-mentioned issue, we have come up with a project that makes video streaming convenient and hassle-free. Now people do not have to crave good network quality anytime. All they need is sufficient data to stream a video as low as 144p resolution and we will take care of the rest.

• Our platform which enhances the low quality (144p) video to its corresponding high quality (360p) video.
• This is done through Super Resolution Deep Convolutional Neural Networks.
• This model takes in a low-quality image and returns its corresponding high-quality enhanced image and this image is continuously sent to the webpage where the user can see the output.
• The entire process requires internet usage as low as streaming a 144

Challenges we ran into

-Extraction of youtube videos without downloading them locally

-Streaming multiple videos at a time

-To build an accurate super resolution deep convolutional neural network

-Since it is a deep learning model, one of the major challenge we overcame was to reduce the model size to as low as 6 mb so that we can provide cutting edge performance with as less lag as possible, which can be further reduced to provide a great user experience