Hats4Cats

NOTE: Please open the PowerPoint link IN POWERPOINT. The video is the second in link in the Try it Out links. The video is included in the presentation, it WILL NOT run on google slides or otherwise. THANK YOU!

Inspiration

We were inspired by Dr. Seuss' Cat in the Hat.

What it does

The app takes a picture of a cat that the user inputs, and outputs the same picture with a hat of the user's choice.

How we built it

A user selects an image of their furry friend along with a pre-made hat, then using the app, makes a POST request to an Nginx webserver. The webserver then passes the input image along with the hat selection to an S3 bucket used for input. A Lambda microservice then detects new files in the bucket. It then takes the input image and passes it through a Neural Network model which detects feline facial features (Eyes, Ears, Nose). The microservice then uses this information to place and rotate the selected hat onto the appropriate place in the image. The output image is then placed in an output S3 bucket. The webserver then responds to the user's request with the output image. The image is then fed back to the user via the application. Garbage clean-up in both the input and output S3 buckets is automatically triggered after nginx indicates a successful TCP teardown.

Arbab helped on the front-end dev, Aaron did front-end and back-end, and Faith did graphic design/UX.

Challenges we ran into

Due to the time constraints, we were unable to fully implement the AWS as was our intention. This was because human facial recognition software cannot accurately detect feline faces, so we would have needed to train our own neural net. Our team was also hybrid, so it was difficult to communicate in a timely manner compared to if we were all in person.

Accomplishments that we're proud of

Even though the AWS was unable to be implemented, we are proud that the python runs and that the facial geometry we calculated was accurate.

What we learned

Aaron and myself learned that human facial recognition does not translate well to feline facial recognition. I myself (Faith) learned more about python, and was able to substantially contribute to a project at a higher level than I am used to as a first-year CS student.