Ali is a mod on Reddit and thought that it would be useful to have a method to reverse image search to see if a comic strip has been re-posted without credit. So we wanted to make a site to allow for this.
What it does
The website allows to search for an image through a database and uses dhash to find the most similar image.
How we built it
We used nodejs with an sqlite server to store the hashes for all the images for xkcd that we then search through to find the comic with the smallest diff to the uploaded image.
Challenges we ran into
Getting the image to upload was an issue. Another challenge was building the database with all the comic strips. Also getting azure to work for this was less than easy.
Accomplishments that we're proud of
Getting it to work.
What we learned
How to use dhash and sqlite to build a reverse lookup image database.
What's next for Coogle
Spice it up and make it public facing.