FarmGuardian
An offline farming assistant. Built for Punjab, with AI that works in the field.
Final Year Project · Built with a 3-person team under faculty supervision.
The Problem
Farmers need crop disease detection and practical farming guidance even when an internet connection is unavailable. FarmGuardian brings these tools together in an offline-first mobile app for Punjab farmers.
What I Built
On-device tomato leaf disease detection using TFLite and EfficientNet-B0, achieving 91% accuracy.
Random Forest yield prediction for maize, wheat, rice, tomato, and potato, paired with cure recommendations.
Crop planning with water and fertilizer guidance, crop recommendations, and daily mandi price tracking.
Full Urdu localization through i18next, with Supabase supporting the application.
Outcome
Won 3rd Place and 1st Place (People’s Choice) at the COMSATS Abbottabad FYP Competition, Spring 2026. The reported 91% accuracy applies specifically to tomato leaf disease.
Tech Stack