How to navigate the streaming wars (using Computer Vision) ?

Thomas Taieb
3 min readNov 30, 2019

Netflix, Hulu, HBO Go, HBO Now, HBO Max, Disney +, Apple TV +, Amazon Prime Video, Youtube TV and I am forgetting a few. You just wanted to Netflix and chill. Now you have so much choice that you can’t even focus on the chilling anymore…

Who didn’t spend at least a half hour trying to decide what to watch? Browsing endlessly the catalog. So much options. Yet so little information about the titles… That you end up “IMDBing” the movies/tv shows before deciding. In the world of Alexa, voice and image activated services, manually doing the process of searching IMDB is just so 2018. There must be a startup to solve that problem and make people’s lives a little (a lot) better!!

Well, not a startup, but an “app”. Though I am open to discuss with any streaming platform about a potential “acquihire”. As long as the “acquisition price” has at least one zero in it, we’re good to go.

The concept is simple. Simplicity is valuable. You aim the camera of your phone at the screen and the IMDB score of the movie/TV shows directly top of the image. You have nothing else to do. You can now make an informed decision and go back to chilling. Or get back to that huge bucket of popcorn which is waiting for you.

“Wow… You must be incredibly good at coding to do something like that”. Said no one, ever. Maybe my mom. Or even she realized it would be too big of a lie. Nop. I just used AWS. They are the best wingmen you can ask for: they do all the heavy lifting but let you get all the praises. For a little bit of money. That’s just good business!

I decomposed the process in three steps:

Step 1: fix the ground rules with your wingman

Your wingman (or wingwoman by the way) needs to know what you are looking for in order to be efficient. Well AWS is the same. And the app needs two things: first, to recognize if there is a TV (true, you could be streaming on a different screen, but as I said, I like simplicity) and if there is, to locate and extract the text within the images (text which would be the title of the movie/TV show — once again, there are no protections against misuse).

Step 2: go talk to the person of interest

It’s not enough to plan and strategize, now you need to actually do something. To give you the rating, the app needs to go get it from IMDB. Unfortunately, there doesn’t seem to be an existing API to easily fetch the data (thank you Amazon — Amazon owns both AWS and IMDB). I could use other data sources, but that would be cheating. Therefore I decided to use Google in order to get the right link. Amazon doesn’t have a search engine (per se) so it is not cheating!

Step 3: scrape, scrape, scrape!

Step 3 is the part where you take whatever you want, and just leave. With absolutely no concerns about the other party involved. As long as the app was upfront from the beginning, it is okay! Kind of. Hopefully. This is the part where the app scrape the HTML link in order to get the rating (meaning reads the whole page and automatically returns exactly what we want).

Now the app just needs to return the information to you, its Dark Lord. Of course, there are many, many, maaaany ways to make it better. But to be fair, my objective was just to do some scraping. And let’s be real, winter is already here (for you, Nothern Hemispherener), so I know that you needed this tool to be functional as soon as possible. You’re welcome!

Below is an example of output. PS: not a fan of this show…

Example of output
Example of output

By the way, I am welcoming any feedback and/or any collaboration/exchange of ideas on existing or future projects that could be fun to build.

As usual, you can find the code here.

--

--

Thomas Taieb

Financial analyst in M&A, love everything about technology, start-ups and business.