Morasho/shamba-ai

# Shamba AI 🌿 Crop disease detection for Kenyan smallholder farmers. Upload or photograph a diseased leaf → get a diagnosis, treatment steps, yield impact estimate, and nearest agro-vet — in English or Swahili. --- ## Quick start (local dev) ```bash cp .env.example .env # fill in your keys docker compose up # API + worker + Redis + Postgres + frontend ``` Frontend: localhost API docs: localhost --- ## Project structure ``` shamba-ai/ ├── backend/ │ ├── app/ │ │ ├── api/routes/ # FastAPI routers │ │ │ ├── diagnosis.py # submit, poll result, feedback │ │ │ └── knowledge.py # disease knowledge base │ │ ├── core/ # config, database │ │ ├── models/ # SQLAlchemy models │ │ ├── schemas/ # Pydantic schemas │ │ ├── services/ │ │ │ ├── preprocessing.py # CLAHE, unsharp mask, white balance │ │ │ ├── inference.py # ONNX Runtime + TTA + temperature scaling │ │ │ ├── gradcam.py # Grad-CAM heatmap generation │ │ │ ├── knowledge.py # disease KB with Kenyan treatments │ │ │ ├── storage.py # Cloudflare R2 │ │ │ └── sms.py # Africa is Talking SMS │ │ └── tasks/ │ │ └── celery_app.py # async inference worker │ └── ml/ │ ├── train.py # EfficientNet-B0 + Albumentations │ ├── export_onnx.py # PyTorch → ONNX │ └── calibrate.py # temperature scaling └── frontend/ └── src/ ├── components/ │ ├── capture/ # CaptureScreen with tips + quality check │ ├── diagnosis/ # Result screen + diagnosing screen │ ├── layout/ # AppLayout + header │ └── ui/ # LanguageToggle ├── hooks/ │ └── useDiagnosis.ts # full flow: compress → quality check → upload → poll ├── lib/ │ ├── api.ts # API client │ ├── imageUtils.ts # compression + blur detection │ └── uploadQueue.ts # IndexedDB offline queue └── i18 …