Inspiration
Being outdoors is a great release from working at a desk job, there are so many animals to uncover and sometimes you don't even know what you have just encountered. It's a good safety measure to check what animal you have seen and if you should take any sort of safety precautions.
What it does
The application accepts two inputs: 1) Image of the animal you are looking to uncover more about 2) Location of the user, some species are native to certain parts of the world and it's important to give this context in order to accurately classify the species.
How we built it
The project was built using solely Python on the heels of FastAPI and Streamlit. The backend uses the Sonar API after detecting the most probable animal species with the use of PyTorch and HuggingFace's Transformers library using the VIT model. The Sonar API is responsible for fetching data about the animal with additional context included with the "web_search_options".
Challenges we ran into
Due to the inclusion of PyTorch and the development on Linux. NVIDIA drivers were needed to include these requirements in the project.
Accomplishments that we're proud of
Having a full-stack interface application that can process user input and give users helpful information!
What we learned
Building a project can be challenging, but very rewarding in terms of seeing the finished product and learning all the different tools necessary to fulfill the vision.
What's next for Animal Safety Checker
I think another helpful use case would be something similar to the original Animal Checker but this time for mushrooms. Mushroom hunting is a big part of my ethnic culture and there are a lot of species of mushroom, however the majority are not edible and it requires serious understanding to pick the safe ones.
Log in or sign up for Devpost to join the conversation.