Skip to content

SergioWinn/Detectomato

Repository files navigation

🍅 Detectomato – A CNN-Powered Tomato Leaf Disease Detection App

Detectomato is a mobile application built with Flutter and Dart to automatically detect diseases on tomato plant leaves. This app leverages a Convolutional Neural Network (CNN) with the ResNet50 architecture, optimized with TensorFlow Lite to run directly on Android devices for offline use.


🔍 Key Features

  • 📸 Capture images or upload photos of tomato leaves
  • 🧠 Automatic prediction of 11 common leaf diseases
  • 📊 View a history of detection results
  • 👤 User authentication (sign up, sign in, sign out)
  • 📝 Support and feedback forms
  • 💾 Backend powered by Supabase

💡 Tech Stack

Technology Description
Flutter Mobile app development framework
Dart Primary programming language
TensorFlow Lite On-device image classification model
Supabase Open-source backend service (PostgreSQL, APIs)
ResNet50 CNN architecture for leaf disease classification
Kaggle Dataset Source of training data for the AI model

🧠 About the AI Model

The CNN model was developed and trained using a dataset from Kaggle:
🔗 Tomato Disease Dataset – Multiple Sources

Recognized Disease Classes:

  • Late Blight
  • Early Blight
  • Septoria Leaf Spot
  • Bacterial Spot
  • Leaf Mold
  • Target Spot
  • Tomato Mosaic Virus
  • Tomato Yellow Leaf Curl Virus
  • Spider Mites
  • Powdery Mildew
  • Healthy

📌 Project Status

  • ✅ Done: Offline AI-based disease detection (ResNet50 + TFLite)
  • ✅ Done: User login, registration, history, and classification results
  • 🟡 In Progress: Adding detailed disease descriptions and treatment recommendations
  • 🟡 Future: Integrating geolocation and push notifications

👥 Tim Pengembang

  • 👨‍💻 Sergio Winnero – AI/CNN Model, Flutter, TensorFlow Lite, Frontend, Backend
  • 🧪 Samuel Setiawan – Testing, Finalization
  • 🎨 Karina Vanya Wardoyo – UI/UX Development

About

Detectomato: An offline-first Flutter app for detecting tomato leaf diseases using a ResNet50-based CNN model optimized with TensorFlow Lite. Features user authentication, detection history, and Supabase integration.

Topics

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors