WildDex

Login page
Profile posts
Landing page
Creating a post - more info
Creating a post
Finished post

Inspiration

Given the topic of wildlife conservation, we reflected on our own experiences of encountering wildlife out in nature. We wanted to capture the excitement of the moment so that it becomes a rewarding memory. Being aware of the difficult task of cataloguing and collecting data on wildlife, we wanted to create a fun app that allows users to record their encounters with wildlife while contributing to data collection.

What it does

The app's experience is supposed to simulate the excitement of discovering Pokemon and indexing them, it is your personal “WildDex” to index the wildlife you encounter.
The app's users will be able to record their encounters with wildlife using their smartphone’s camera and save the location the sighting occurred at.
The users will be able to crowdsource data on wildlife habits by tracking their encounters, which can then be useful for conservationists to seek patterns on how certain wildlife is affected by climate change and human actions.
The wildlife will then be identified by an image classification model that uses machine learning.
Animals that the user captures will be posted under their account and be available to view on a community feed.
The user will receive special badges based on interaction on their posts in the community feed through upvotes.

How we built it

Secure user authentication with Clerk
Mobile application built with React
PostgreSQL for the app database storage
Machine Learning models trained with TensorFlow
Supabase used as a bucket for image storage and retrieval
Google Maps API integration for geo-tagging
Docker for build containers
API server created with Express

Challenges we ran into

Image classification becomes difficult when there are many classes to identify.
Creating a neural network architecture to correctly classify images of animals took a lot of fine-tuning and research.
Hosting the machine learning model in a server so that it can be used alongside a web interface.

Accomplishments that we're proud of

We have a very cohesive and easy-to-use interface that allows users to quickly snap a photo and geo-tag their capture.
Integration with Google Map's API.
Machine learning applications.

What we learned

Power of teamwork
Discovered integrating new technology, such as Google Maps API, and authentication with Clerk.

What's next for WildDex

We would like to create a way for authorized users to access the database of animal sightings for research.
Expand animal classification capabilities to include more animals and geographical regions.