🤩Getting Inspired by the Netflix ML Team

ML is all about probem-solving

In my opinion, any technology has more to it than just making our lives easier. It is about how you identify & perceive a problem and how you go about solving it. It has the potential to bring radical changes when put to right use. ChatGPT, taking a recent example, is case in point.

The desire to bring/be a part of such radical changes through ML is what inspires me to not vanish & go open a small bookstore in a foreign country and spend the rest of my life as an anonymous quirky outsider who has a lot of cats. (too specific?)

Anywho, So back to the magic that the Netflix ML Team made happen!

Easy visual content search system

People at Netflix observed that whenever the content creators/editors wanted to produce an artwork or a trailer, most of their time was occupied in scrubbing for frames, scenes of specific characters, events, attributes (like close-up shots) etc.. And it is one tedious task when you have a large content library (I mean, Netflix c'mon). Here is an example of a trailer the editors created using such methods.

The usual number of frames per second in a 120 min theatrical movie is 24. So there are 24 x 60 x 120 = 17,280 frames in a typical two hour movie. And if you feel like doing the math for total number of frames in the whole wide Netflix content library right this moment, there are 3,600+ movies and 1,800+ TV shows streaming on their platform (when I wrote this blog).

In other words, editing/creating was a pain.

So, they came up with a really cool visual element search system for their content creators/editors where the user could just enter what they are looking for in natural language and the system will return all the frames with the requested visual element. For example, here is what the system returns when the user searches for red race car-

A very innovative and helpful solution, isn’t it? To read more, Here’s the blog: Building a Media Understanding Platform for ML Innovations .

So, the next time maybe when you are wondering about a cool project to work on take a look around? YOU NEVER KNOW what use case could pop up.

My intention for this blog is to just share some really cool & innovative ML solutions thought up by some really cool & innovative people. Observing that, we can learn How to Think, Where to find motivation and How to make it happen. So that, any of us can create/be a part of such use cases some day!