LeafSafe | Devpost

Inspiration

As an avid hiker and outdoor enthusiast, I’ve always been passionate about nature and technology. I wanted to bring these two passions together to help fellow outdoor explorers stay safe by identifying plants in the wild.

What it does

LeafSafe uses computer vision and a LLM to identify whether a plant is edible or dangerous through a simple photo. It also provides first-aid advice for harmful exposure and details about local poisonous plants based on location and season.

How we built it

We used ChatGPT-4 for computer vision to identify plants, and ChatGPT-3.5 as our large language model (LLM) to provide advice. For our full-stack framework, we chose Next.js with Tailwind CSS for styling, and Node.js for backend logic. We manually deployed the code using AWS EC2, with AWS Application Load Balancer enabling scalable deployments, and nginx for handling proxy requests.

Challenges we ran into

One of the main challenges was ensuring the AI model's accuracy with different lighting conditions and angles. Additionally, curating a comprehensive database of plants and creating an intuitive UI required careful design.

Accomplishments that we're proud of

We’re proud of creating a tool that not only helps outdoor enthusiasts but also empowers them with essential safety information. The AI’s ability to recognize various plants with high accuracy is a significant accomplishment.

What we learned

Throughout the process, we learned how to optimize image recognition models for real-world scenarios and how to combine user-friendly design with essential safety features. This balance was key to making the app useful and accessible.

What's next for LeafSafe

We plan to expand the plant database, improve AI accuracy, and add offline functionality for remote areas. We also want to include real-time tracking of local plant dangers based on seasonal changes.