This project is an AI-powered web platform that automates plant disease detection and streamlines agronomic decision-making. Built with deep learning, it features a dual-stage pipeline: a binary EfficientNet-B0 model determines if a leaf is healthy or diseased, and an infected sample is then routed to a multi-class EfficientNet-B3 model to identify the specific condition across 26 distinct disease categories.
The backend is developed using Flask, integrating automated data engineering to sanitize and map classifications directly to an Excel/CSV reference sheet. This live mapping replaces raw model indices with descriptive common names and pulls comprehensive clinical fields, including primary/secondary causes, climate management vulnerabilities, and active chemical treatment strategies.
The single-page web interface focuses on an interactive user experience with seamless typography. When a crop leaf photo is dropped or uploaded, the file box dynamically swaps into a secure image preview showing successful upload feedback alongside a quick reset button for re-evaluation. The results are instantly rendered in a structured grid, providing farmers and agronomists with clear, actionable insights to protect crop yields.
Log in or sign up for Devpost to join the conversation.