TomatoScan

TomatoScan puts a plant-disease expert in the farmer's pocket. A farmer photographs a tomato leaf and uploads it; a TensorFlow/Keras model served through Flask classifies the 512x512 image and returns the diagnosis, a confidence score, the symptoms observed and treatment recommendations. Below 60% confidence it says the leaf is not recognised instead of guessing. Every scan is saved, with a PDF diagnostic report, a treatment log that links follow-up scans, weather data, and an outbreak map of anonymised GPS-tagged scans. It is designed for poor connections: it installs to the phone's home screen, caches its pages for offline use and switches to Swahili in one tap. Built with PHP, MySQL, Flask and TensorFlow, and also packaged as an Android app.