Imagerec

Inspiration

Many tag-based image boards already exist. But they all lack one certain feature: Image recommendations. If we like certain kinds of images, we want to see similar images.

What it does

The core is an image board, ie. a massive database of images that are organized by tags. You can search tags (eg. "bird" or "landscape") and find images that are tagged as such. As the user naturally browses images, they will accumulate a list of tags. The recommendation algorithm will take these tags into account and serve the user images that contain tags that they frequently see.

How we built it

The backend was written in Python, and the GUI was written in Qt6. Our sample image downloader was written in Javascript and made use of Pixabay's API, taking advantage of the royalty-free licence on their images for our demo.

Challenges we ran into

Qt6 was a brand new toolkit for us, and we had to learn it on the fly. Our team is quite inexperienced in the ways of frontend design, so learning Qt ended up being quite the bottleneck for us.

Accomplishments that we're proud of

Writing an image database from scratch, with data structures to support images with tags as well as users with their images viewed and tag frequencies.