Potato Leaf Disease Detection App

This application detects potato leaf diseases (Early Blight, Late Blight, or Healthy) using a trained deep learning model. It allows users to upload images, capture live images via webcam, or analyze real-time video streams for disease prediction. The app also provides visual Grad-CAM heatmaps to highlight areas of focus for predictions.

---

Features

1. Upload Image

   • Upload a potato leaf image for classification.
   • Displays predictions from three models: Custom CNN, Inception, and ResNet.
   • Shows Grad-CAM heatmap overlay for better interpretability.

2. Use Camera

   • Capture images directly using a webcam for classification.
   • Predicts disease type and displays model confidence.

3. Real-Time Video Analysis

   • Analyze live webcam video streams to detect diseases.
   • Shows bounding boxes dynamically focusing on activated regions.
   • Confirms prediction after analyzing 15 frames for stability.

4. About Section

   • Detailed description of each feature and models used in the app.

---

Models Used

1. Custom CNN Model: Lightweight model designed for quick predictions.
2. Inception Model: Pre-trained model fine-tuned for potato leaf disease classification.
3. ResNet Model: Another pre-trained model for robust classification.

---