Have you ever played Pokemon Snap?
We wanted to create a quick and easy way to identify birds in the wild.
What it does
We’ve built a full-stack application that identifies birds from your camera. You just find the birds–we’ll identify their common name, scientific name, and a quick description. It works right through your device’s camera to capture a snapshot of any birds in the vicinity.
How we built it
It works right through your device’s camera to capture a snapshot of any birds in the vicinity. We accept the request using a flask server, which passes the appropriate information to our microservices. Using Google Cloud’s Vision API, we can identify bounding boxes on all the birds. We crop these images and send them to a tensorflow model that classifies the species of the bird in the given image. Once we’ve identified a bird’s species, we can use Wikipedia’s API and parse information from any given bird’s page. All this info is sent back to the frontend, where we display all the information for all birds in the original photo sent.
Challenges we ran into
We used a variety of different APIs and libraries, many that we've never worked with before. It was difficult figuring out how to glue everything together.
Accomplishments that we're proud of
There were a lot of firsts for us in this hackathon. For one member, it was their first time writing in Python. For multiple of us, it was our first time working with APIs.
What we learned
We learned a lot about how to integrate APIs into a project!
What's next for Who's that Birdy-mon?
We'd like to expand the app to work for more than just birds, such as other animals or even inanimate objects.