code-cheff/E-Mali

E-Mali is an plant disease detection app which works on self trained model and is backed by openrouter.ai api for detailed solution for plant diseases. # 🌿 Plant Disease Identification — Android App An Android app that identifies plant diseases from a photo of a leaf, running a **Convolutional Neural Network (transfer learning, MobileNetV2)** **on-device** with **TensorFlow Lite** — no internet connection required at prediction time. > BCA Final Project (DCA3202) — *Plant Diseases Identification* > Stack as per synopsis: Python · TensorFlow/Keras → **TensorFlow Lite** · CNN · > OpenCV/PIL · NumPy/Pandas · Matplotlib · **Android (Kotlin)** · Git --- ## ✨ Features 1. **Capture or pick** a leaf photo (**Camera** or **Gallery**). 2. **On-device CNN inference** via TensorFlow Lite — fully offline, no internet needed. 3. **Rich result card:** predicted disease, crop name, color-coded **Healthy / Diseased** status chip, and an animated confidence bar. 4. **Disease knowledge base:** a plain-language **description** and **recommended treatment/action** for each disease (bundled offline). 5. **Top-3 possibilities** shown as visual confidence bars. 6. **On-device inference time** displayed (e.g. "Analyzed on-device in 38 ms"). 7. **Low-confidence warning** prompting a clearer photo. 8. **Share** the diagnosis as text to any app. 9. **Session scan history** of the last 5 scans with thumbnails. 10. **Material 3 UI** — app bar, cards, tonal/outlined buttons, progress indicators; background inference keeps the UI responsive. There are **two parts**: | Part | Where | Language | Role | |------|-------|----------|------| | **Model training** | project root (`train.py`, `src/`) | Python | Train the CNN once, export a `.tflite` model | | **Android app** | `android-app/` | Kotlin | Ship the `.tflite` model and run it on the phone | --- ## 📁 Project structure ``` plantDisease/ ├── train.py # Trains the CNN and exports model.tflite + labels.txt ├── make_sample_dataset.py # Tiny synthetic dataset for smoke-testing the pipeline ├── requirements.txt # Python (training) dependencies ├── src/ │ ├── config …