PicGenie in action, scanning a cute cat picture.
No one knows everything. Luckily for us, we live in the age of information. Unfortunately, it often seems that the hardest thing about learning is finding out what to ask. With PicGenie, users are liberated from needing to luckily chance upon the correct keywords, and instead search whatever they want directly through pictures of the world.
What it does
PicGenie allows users to text a picture to 508-762-4753, and receive back a guess and short wikipedia description of what that picture is thought to be. The searches are then displayed in a ring graph data visualization.
How we built it
We used twilio to receive the pictures, clarifai to get the most pertinent tag, the wikipedia api to retrieve the tag word's description, Charts.js to create a data visualization, and Flask to provide the infrastructure
Challenges we ran into
It was quite challenging to get so many API services to work together nicely. In particular, using Flask to simultaneous affect both the returning messages and the data visualization was tricky to implement.
Accomplishments that we're proud of
We are proud of getting a project with so many moving parts off the ground.
What we learned
We learned that in a project like this, reliable progress can only be made when work is partitioned effectively.
What's next for PicGenie
We want to improve our parsing of the clarifai json that was returned, so that we can provide better image recognition. In addition, our data visualization will be made more dynamic and accessible to the end user.