How harshayalagani29/crop-disease-ai works
This MVP lets a farmer submit a leaf image and location, then returns a disease classification, risk assessment, recommendation, and optional weather context; a history view retrieves prior predictions. The API coordinates inference, weather lookup, scoring, guidance, and SQLite persistence. Separately, the ML scripts prepare data, train the MobileNetV2 model, and evaluate it. The README documents those capabilities, while the shown API and UI excerpts provide direct evidence for the main runtime calls; some training-to-runtime links are documented rather than visible in source.
Main entry points: main.py FastAPI endpoints · App.jsx Detection and history UI · train.py MobileNetV2 training · split_dataset.py Dataset splitting
Components 14
Farmer interface
- Detection and history UI
frontend/src/App.jsx - English and Kannada
frontend/src/i18n.js - React entry
frontend/src/main.jsx
Decision API
- FastAPI endpoints
backend/main.py - Image inference
backend/model_service.py - Risk scoring
backend/risk_engine.py - Recommendations
backend/recommendations.py
Persistence and services
- Weather API
- SQLite database
backend/db.py
Model lifecycle
- Class extraction
ml/extract_classes.py - Dataset splitting
ml/split_dataset.py - MobileNetV2 training
ml/train.py - Model evaluation
ml/evaluate.py
Other components
- Farmer
Connections 13
- Farmer → Detection and history UI: uses
README.md - React entry → Detection and history UI: renders
frontend/src/main.jsx - Detection and history UI → English and Kannada: uses labels
frontend/src/App.jsx - Detection and history UI → FastAPI endpoints: requests predictions and history
frontend/src/App.jsx - FastAPI endpoints → Image inference: runs inference
backend/main.py - FastAPI endpoints → Weather API: fetches weather
backend/main.py - FastAPI endpoints → Risk scoring: assesses risk
backend/main.py - FastAPI endpoints → Recommendations: gets recommendation
backend/main.py - FastAPI endpoints → SQLite database: initializes and reads/writes
backend/main.py - Class extraction → Dataset splitting: prepares source images
- Dataset splitting → MobileNetV2 training: prepares training data
README.md - MobileNetV2 training → Model evaluation: provides trained model
README.md - MobileNetV2 training → Image inference: supplies model for inference
README.md