# 🌿 TouchGrass AI Bird Explorer
Discover nature. Identify birdsong. Keep a digital nature journal.
What if AI could encourage us to spend less time scrolling and more time exploring the world around us?
TouchGrass AI Bird Explorer is an AI-powered project built on Arduino UNO Q that helps people explore nature through bird sound identification and AI-assisted journaling.
The idea is simple: use technology to bring people closer to nature, not keep them glued to a screen.
🐦 What I Built
I built a web application with two main features:
1. Bird Sound Identification
Upload a bird audio recording and use BirdNET to get possible bird species predictions. The results help users investigate the sounds they hear outdoors.
2. AI Nature Journal
Write a personal observation about something you saw or heard in nature. A locally hosted Qwen2.5 language model helps turn the observation into readable journal notes.
The journal is designed to organize what users report, not replace their own observations. AI-generated notes should be reviewed for accuracy.
🛠️ How I Built It
The project combines:
- Hardware: Arduino UNO Q
- Bird identification: BirdNET
- AI journaling: Qwen2.5-0.5B-Instruct
- Local model serving: llama.cpp
- Backend: Python and Flask
- Frontend: HTML, CSS and JavaScript
The Flask application connects the web interface with the bird-analysis and journaling workflows. The Qwen model runs through a local llama.cpp server.
🌱 Why TouchGrass?
Technology can help us learn about the world beyond our screens.
This project explores how AI can encourage people to listen carefully, notice wildlife, and record their outdoor experiences. It combines practical AI experimentation with environmental curiosity.
🧩 Challenges and Learnings
Connecting the web application to a locally hosted language model required troubleshooting the model server and its connection to the application.
This is a working prototype. Voice input reliability, deployment, and additional testing remain areas for improvement.
🚀 What's Next?
- Improve voice input and speech-to-text support.
- Add searchable observation history.
- Improve mobile usability.
- Test with a wider variety of bird recordings.
- Explore ways to organize observations over time.
🔗 Source Code
GitHub repository:
https://github.com/LaxmikanthJanardhan01/touchgrass-bird-explorer
The project currently runs locally on Arduino UNO Q; a public live demo is not available yet.
🌍 Open Innovation
Open-source tools give students and developers the opportunity to experiment, learn, and build on existing technology.
TouchGrass AI Bird Explorer brings together open tools, local AI inference, and accessible hardware to explore a simple idea: technology can be a companion to curiosity about the natural world.
Built with curiosity, AI, and a love for nature. 🌿🐦
Top comments (0)