Nature Explorer: A Local AI Field Journal

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass What I Built Nature Explorer is a digital field journal designed to help people notice more when they step outside. Take a photo of a plant, bird, insect, animal, fungus, or natural feature. Choose a category, and the field assistant suggests an identification, description, likely habitat, confidence level, and a short field note. You can save the discovery to a personal collection, then search and filter your finds later. The collection stays in the browser, so there is no account to create and no database to set up. The idea is to make the screen a quick companion for an outdoor observation-not the destination. Demo Code https://github.com/suryastack/Nature-Explorer How I Built It The frontend is built with React, TypeScript, and Tailwind CSS. A separate Node.js and Express backend accepts the image and sends it to Ollama, which runs the open-weight Gemma 3 4B vision model locally. The model returns a structured suggestion that the app displays as a field note. Discoveries are saved in the browser: image files go into IndexedDB, while lightweight details such as category and date are stored in localStorage. There is no hosted AI API, database, or user account in the MVP. After downloading the model, image analysis runs through Ollama on the user's computer. Why Does Open Innovation Matter? Nature photos can reveal personal routines and locations. Running the vision model locally lets people explore an image without sending it to a hosted AI service or creating an API account. Open-weight models also make the project easier to adapt. Developers can inspect the setup, experiment with other compatible vision models, and improve the field assistant without redesigning the whole app around a closed API. For this project, open innovation made a private, locally run nature journal possible on an everyday laptop. Prize Categories Best Use of Gemma Top comments (0)