🔗 Click here to watch the demo video
Welcome to the TomatoLeaf Care. This tool uses deep learning to spot and classify diseases in tomato leaves 🌱. Made for farmers 👩🌾, it helps them make smarter decisions about their crops. The app currently identifies 9 tomato leaf diseases:
1️⃣ Tomato Mosaic Virus
2️⃣ Target Spot
3️⃣ Bacterial Spot
4️⃣ Tomato Yellow Leaf Curl Virus
5️⃣ Late Blight
6️⃣ Leaf Mold
7️⃣ Early Blight
8️⃣ Spider Mites (Two-Spotted Spider Mite)
9️⃣ Septoria Leaf Spot
Seven deep learning models were trained over 30 epochs to perform accurate tomato leaf disease classification 🚀. The performance of each model is summarized below:
| 🤖 Model | ✅ Training Accuracy | 📈 Testing Accuracy |
|---|---|---|
| 🧩 Novel CNN | 94.52% | 93.00% |
| 📱 MobileNetV2 | 96.78% | 91.69% |
| 🖼️ VGG19 | 97.31% | 89.60% |
| 🏆 ResNet50 | 99.11% | 94.59% |
| 🌐 InceptionV3 | 89.42% | 83.99% |
| ⚡ AlexNet | 97.13% | 94.40% |
| 🔗 Ensemble Model | 98.94% | 94.40% |
🏆 ResNet50 emerged as the best-performing model, achieving:
- ✅ Training Accuracy: 99.11%
- 📈 Testing Accuracy: 94.59%
The project employs FastAPI to connect the deep learning models with the frontend, enabling real-time disease prediction 🔄.
The ExpressJS API is utilized for user authentication 🔒 and data management 📂. It stores user details, prediction results, and feedback in a MongoDB database.
The ReactJS-powered frontend offers users a seamless experience ✨. Key features include:
- 🔮 Predict Page: Upload a tomato leaf image and get the predicted disease.
- 📚 Resource Section: Learn about various tomato diseases in detail.
- 👤 Profile Section: View the user's profile picture, full name, and email address.
- 🕒 History Page: Track all prediction records.
- ℹ️ About Us Page: Learn about the project and provide feedback via the "Give Feedback" button.
The app checks whether the uploaded image is a leaf 🌿 using an InceptionV2 model trained on a dataset combining tomato leaf and CIFAR-10 images. If the image isn't a leaf, an error toast notifies the user 🚫.
The project is deployed across multiple platforms:
- Frontend: Vercel
- ExpressJS API: Render
- FastAPI API: Hugging Face Spaces
1️⃣ Live Preview: https://tomatoleaf-care.vercel.app/
2️⃣ FastAPI Deep Learning Model API: https://som11-multimodel-tomato-disease-classification-t-1303b88.hf.space/
3️⃣ Swagger Documentation (Deep Learning Models): https://som11-multimodel-tomato-disease-classification-t-1303b88.hf.space/docs
4️⃣ FastAPI Leaf Validation API: https://som11-tomato-leaf-or-not-tomato-leaf.hf.space/
5️⃣ Swagger Documentation (Leaf Validation): https://som11-tomato-leaf-or-not-tomato-leaf.hf.space/docs
6️⃣ ExpressJS API: https://tomato-leaf-classification-project-10.onrender.com/
Although the models achieve high accuracy 🏆, they may occasionally misclassify diseases or fail to detect them entirely. Do not rely solely on the model's output. Use it as an aid, not as the sole source of decision-making. 🌟