AI-paca

Sample image of animals. Taken literally 5 hours ago from submission time.
Visualization map of animal distributions.
Homepage

Inspiration

We were inspired from a class that Lawrence is taking, AI For Conservation. His professor conducts research on deep learning tasks. His work is reliant on datasets collected from photographers who go out into outdoor environments and take pictures of the surrounding wildlife. The big idea for this project was founded upon the question: What if we could leverage a wilder demographic of people to help source this data?

What it does

Users can use the AI-paca mobile application to take pictures of animal sightings. When they submit entries, the images get analyzed by a deep learning classifier to automatically identify the species of the animal(s). This data is consolidated and researchers can access, visualize, and download the data through a website interface. In short, AI-paca allows mobile users to contribute to environmental conservation efforts through a mobile app and gives researchers access to a larger pool of crucial ecological data.

How we built it

React and React Native were used to implement our web and mobile apps respectively. Since they are both in javascript, we can implement our frontend architecture all in one language. In addition, the AI model and the api are both running on python frameworks, making them directly inter-operable. Finally, we used supabase because it provides a cloud-hosted database along with a managed file storage - all for free. We deployed the web infrastructure on an AWS EC2 instance with GPU support. By using these technologies we were able to minimize technical overhead when implementing our application.

Challenges we ran into

We were all very busy with our commitments this weekend. As a result, we couldn’t spend as much time as we wanted on AI-paca. Everyone was also very specialized with a portion of the project so we all had to spend time familiarizing ourselves with the whole stack. Because of this focus, we ran into some trouble making sure the frontend and backend worked with each other.

Accomplishments that we're proud of

This is our first hackathon and we feel very proud of what we accomplished in the limited time given. And although we didn’t get all the features we initially planned on implementing, we are still really satisfied with the outcome of the project.

What we learned

We learned to work together as a team and for everyone to do their part since each team member was tasked to complete a portion of the project, splitting it up into the website, mobile, database, and AI implementation. The technology we worked with was also very interesting and we all enjoyed what we did.

What's next for AI-paca

We hope to first provide the project to Lawrence's professor and see what he thinks the future of AI-paca could look like. We would love to continue development and hopefully release this for public use by adolescents and even adults. We want to protect our environment and ecosystem and this is one way that we believe we can help out. We would like to also adjust the application so that it can better uphold FAIR and CARE principles of conservation. This application has the power to significantly benefit our environment. However, we need to ensure that it is doing so in a sustainable and ethical manner.