PlantGuard is an intelligent mobile-based plant disease detection and treatment system developed to assist farmers and gardeners with accurate, fast, and actionable plant health diagnosis using deep learning. The system leverages a ResNet-18 deep learning model trained in PyTorch on a well-curated multi-class dataset (with up to 8 disease classes per crop) to identify diseases from leaf images.

The model achieves strong performance even on modest datasets (≈500 images per class), and is optimized for real-world use and eventual deployment on mobile devices via TensorFlow Lite (TFLite) conversion.

Users interact with PlantGuard through an Android application written in Kotlin, where they can either capture a leaf photo with the device camera or select an image from the gallery. Once an image is submitted, the system classifies the leaf into one of the supported disease categories (e.g., for cotton, maze, potato, etc.) and returns a clear diagnosis.

Beyond detection, PlantGuard provides a comprehensive treatment recommendation system backed by a cloud-hosted database (managed via Supabase). For each disease class, users receive chemical treatment guidance (e.g., recommended fungicides or pesticides with proper use instructions), bridging the gap between AI diagnosis and actionable next steps.

PlantGuard’s goal is to reduce crop losses, improve agricultural productivity, and empower smallholder farmers with AI-driven plant health insights — all from the convenience of a smartphone.