About EcoNova
EcoNova organizes recycling initiatives involving plastic, paper, and e-waste. Through 15+ recycling drives, volunteers have collected 750+ kg of material across multiple schools.
The Waste Composition Analyzer was created to explore how computer vision can support environmental data collection and analysis. It uses a multimodal vision-language model (Google Gemini) to estimate the visible material composition of waste photographs — not to replace physical sorting or weighing.
How it works
- You upload a photograph of a waste or recycling collection.
- The image is validated, resized if needed, and sent to Google Gemini with a structured analysis prompt.
- The model returns JSON describing estimated composition, detected objects, uncertainties, and recommendations.
- The application validates and normalizes the response, then stores the analysis locally.
What this tool cannot do
- Measure exact weight or mass from a photograph
- Detect hidden or obscured materials reliably
- Determine chemical composition or contamination levels
- Replace manual sorting, weighing, or lab analysis
Privacy
Images are stored in a local SQLite database for analysis history. When a Gemini API key is configured, images are sent to Google's API for processing. Do not upload images containing identifiable people or sensitive locations. You can delete individual analyses at any time.