How user-hk-1/cropshield-ai works
CropShield AI is a Streamlit app for identifying a pesticide from an uploaded image and generating crop-specific guidance. A farmer enters a crop and uploads an image; the interface invokes image classification, then requests an advisory report and can submit follow-up questions. The classifier loads its Keras model and labels at runtime, while both language-model features use Groq. The README is empty, so capabilities and connections here are based on the sampled source; asset behavior is shown only where the classifier references those files.
Main entry points: main_app_stm.py CropShield Dashboard · predict_TM.py Pesticide Classifier · llm_app.py Advisory Report · keras_model.h5 Keras Model
Components 8
Farmer Interface
- CropShield Dashboard
main_app_stm.py
Image Classification
- Pesticide Classifier
predict_TM.py - Keras Model
keras_model.h5 - Class Labels
labels.txt
Agricultural Guidance
- Advisory Report
llm_app.py - Follow-up Answers
llm_app.py
Other components
- Farmer
- Groq LLM
Connections 8
- Farmer → CropShield Dashboard: enters crop, uploads image
main_app_stm.py - CropShield Dashboard → Pesticide Classifier: requests prediction
main_app_stm.py - Pesticide Classifier → Keras Model: loads model
predict_TM.py - Pesticide Classifier → Class Labels: reads labels
predict_TM.py - CropShield Dashboard → Advisory Report: requests report
main_app_stm.py - CropShield Dashboard → Follow-up Answers: requests answers
main_app_stm.py - Advisory Report → Groq LLM: generates guidance
llm_app.py - Follow-up Answers → Groq LLM: answers questions
llm_app.py