The intention of repo is to test my skills using FastAI and Streamlit. I built a Bird detection model using CalTech 200-2011 dataset.
This dataset consists of 200 different north american bird species with a total of 11,700 images. The final outcome is to get up 90% prediction rate but am not there yet. The current model has 80% prediction rate which is pretty good using simple FastAI technique.
I built a simple web app using Streamlit framework to test & play around with the model and used Heroku to host and upload the app.
The web app is very simple, just upload any north american birth and it predict the species of the bird along with a next highest probablity of the species
Feel free to play around the project which I hosted here https://classify-birds-app.herokuapp.com/
I plan on uploading the jupyter notebook for the FastAI prediction model in my repo soon.