-
Twitter snapshot of the fake news
Inspiration
The inspiration for this project was first of all the pitch from Lego that talked about the case at the opening. It would be a nice and different challenge to apply machine learning on.
What it does
The focus have been on minimizing the negative impact of leaks. It does this by spreading "Fake news" where the Fake news is mutations of the original leak.The reason it does this is to make the leak more untrustworthy by making it just "one more rumored model", beside all of the mutated leaks.
How I built it
It was build with python, where it starts by crawling the web for relevent images. When a image have been found, will it come through two different image-recognition models. First a fast one, that will look if it's usefull enougth to do the full analysis on. If this is the case, will the full model be used on the picture. The idea for the full model is taken from domain adaption, where we use 3d-models of the things that we want to find, instead of real pictures. If the picture is seen as a leak, will the original 3d-model for the leak be mutation through another model. These mutations is now postet to twitter through the twitterAPI.
Challenges I ran into
Most of the things was new for me, so alot of challenges. But I decided to make my own 3d model files and graphical rendering. This turned out to be a way bigger task than expected..
Accomplishments that I'm proud of
Really the hole project. A bit sad that I didn't have more time, but some new nice knowledge!
What I learned
I have learned webcrawling, 3d-rendering, image-analyzis, post to twitter with the TwitterAPI & Mutation of 3d models
What's next for Fake Lego News
Probably a upgraded version just to satisfy myself
Log in or sign up for Devpost to join the conversation.