GooseBot

Training the model
Training Results
Training Data
Goose!!
Not a goose
0% goose D:

Inspiration

Everybody knows that Geese are the most majestic creature. Why wouldn't you want an AI Discord bot that recognizes the level of Goose within a picture? honk honk

We had this project marked down for a year or so, and this hackathon was the perfect chance to go ahead and create it, just for the sake of it! :>

What it does

Using our custom model trained on 6000+ images, our goose bot classified what percentage of goose an image is, interacting with users in any Discord server.

How we built it

The model was built and trained using tensorflow and keras. The Discord bot uses the Discord API. To plot training data in the testing phases matplotlib and numpy were used.

Challenges we ran into

Our bot initially did not have a class of Not Goose images to compare itself to. This led to errors of every plain image being recognised as a majestic goose. HONK!

Accomplishments that we're proud of

We're proud of learning new skills on a strict time limit. And obviously of creating something to spread our great goose overlord's message.

What we learned

Both of us have separately learnt the main two aspects of this project, the Discord API and python ML. So this project allowed us to "exchange" skills in a way, and allowed us to collaborate and teach other!