Fruit Disease Detection Android App-On-Device CNN (TF Lite) with Remedies

Project Objective

Build a fast, reliable plant disease detector android experience that works even with limited connectivity. Farmers capture a photo, see likely diseases, and follow actionable remedies with safety notes and prevention schedules.

How It Works

1. Capture/Upload: Take a photo of the leaf/fruit or select from gallery.
2. Preprocess: Resize/normalize and feed into a TensorFlow Lite CNN model on-device.
3. Classify: Get top-K disease predictions with confidence scores.
4. Guide: Show treatment steps, chemical/bio control tips, and prevention calendars.
5. Learn: Watch agronomy videos and save notes; sync history to Firebase.

Project Modules

Android Integration Sketch (Java/XML)

// PSEUDO-CODE (illustrative only; Firebase JSON not printed)

// 1) Load TF Lite Model
class TFLiteClassifier {
  private MappedByteBuffer model;
  private Interpreter interpreter;
  private List labels;

  void init(AssetManager am){
    model = FileUtil.loadMappedFile(context, "fruit_model.tflite");
    Interpreter.Options opts = new Interpreter.Options();
    opts.setNumThreads(4);
    interpreter = new Interpreter(model, opts);
    labels = FileUtil.loadLabels(am, "labels.txt");
  }

  Result classify(Bitmap bmp){
    // preprocess: centerCrop -> resize(224x224) -> normalize [0,1]
    float[][][][] input = preprocess(bmp);
    float[][] output = new float[1][labels.size()];
    interpreter.run(input, output);
    return topK(output[0], labels, 3); // top-3 with confidence
  }
}

// 2) Camera/Gallery Flow
void onImageSelected(Uri uri){
  Bitmap bmp = ImageDecoder.decodeBitmap(contentResolver, uri);
  Result r = classifier.classify(bmp);
  viewModel.setPrediction(r); // LiveData -> UI cards
}

// 3) Remedies & Tips
LiveData getTips(String diseaseKey, String fruitType){
  // read from /tips/{fruitType}/{diseaseKey}
  // includes treatment steps, dosage, PPE/safety, prevention calendar
}

// 4) History & Notes (MVVM)
class HistoryRepo {
  void saveScan(Result r, Uri imageUri, long ts){ /* write user-owned scan */ }
  LiveData> list(){ /* realtime listener for history */ }
  void addNote(String scanId, String note){ /* append note */ }
}

// 5) Videos Module
void openVideo(String ytId){
  startActivity(new Intent(Intent.ACTION_VIEW,
    Uri.parse("https://www.youtube.com/watch?v=" + ytId)));
}

// 6) i18n & Offline
// Strings.xml for languages; cache tips; run inference offline; sync when online.

// 7) Safety Notice
// Show PPE & local guidelines banner when treatments are displayed.
              

What You Get