Inspiration
We wanted to help people build greener, healthier spaces — even in tiny apartments or unfamiliar climates. Many plant lovers struggle to choose the right plants for their rooms, especially with pets, limited sunlight, or no gardening experience. We aimed to build something smart, intuitive, and helpful — like a plant whisperer in your pocket.
What it does
PlantPal is an intelligent plant assistant that:
Analyzes room images to detect lighting conditions, window proximity, and temperature
Recommends the best indoor or outdoor plants based on the user’s environment, maintenance preferences, and pet safety
Offers health checkups for existing plants using AI-based image analysis
Tracks care history and sends reminders for watering and fertilizing
Works on mobile via a web interface and a packaged Android app
How we built it
Frontend: Streamlit for rapid UI development and mobile responsiveness
Computer Vision: OpenCV + PIL to analyze brightness, surface zones, and light temperature
LLM: Integrated Ollama (Gemma 3B) for generating personalized recommendations and health diagnostics
Backend: PostgreSQL for storing plant care data, reminders, and user images
Weather API: Open-Meteo + IPInfo for location-based outdoor advice
Mobile App: Android WebView wrapper that loads the hosted Streamlit app as a native APK
Challenges we ran into
Detecting meaningful surface zones and brightness levels from diverse room images
Integrating image uploads and AI prompts dynamically in Streamlit and Ollama
Ensuring mobile compatibility and UI responsiveness under tight time constraints
Packaging a fully functional Android app in 12 hours with a working database backend
Accomplishments that we're proud of
Built an end-to-end working prototype with CV, AI, weather APIs, database, and mobile packaging
Generated real-time, pet-safe plant suggestions from photos
Created a user-friendly, visually clean interface with thoughtful UX
Built a self-contained Android app with health tracking and image history
Deployed a working demo with a reliable backend and live data
What we learned
How to use image-based lighting detection for environment classification
Rapid integration of Ollama's LLMs for real-time visual prompts
Building and deploying a WebView Android app efficiently
Using PostgreSQL to manage reminder logic and image uploads
How powerful even a small model can be with smart prompting and context
What's next for PlantPal
Adding offline support using quantized vision models
Training a custom plant health classifier for real-time feedback
Connecting to smart home sensors for automated recommendations
Allowing community sharing of plant setups and progress
Publishing on the Play Store and scaling to multilingual users
Log in or sign up for Devpost to join the conversation.