watermelon-elite/aaagriculture-models

AAAgriculture ML Models

This repository contains the trained models for the AAAgriculture precision agriculture platform.

Models Included

1. PlantDiseaseModel_best.pth (81 MB)

   • Architecture: ResNet34
   • Task: Plant disease classification
   • Classes: 38 diseases across 14 crop types
   • Accuracy: 95%+

2. crop_yield_model_best.pth

   • Architecture: Custom Neural Network
   • Task: Crop yield prediction (regression)
   • Features: 8 input features (soil, weather, farming practices)
   • R² Score: 0.92+

3. preprocessing_objects.pkl

   • Contains: Label encoders and scalers for crop yield model
   • Used for: Feature preprocessing

4. soil_type_model.pth (130 MB)

   • Architecture: ResNet18
   • Task: Soil type classification
   • Classes: 6 soil types (Sandy, Clay, Loamy, Silt, Peat, Chalky)
   • Accuracy: 90%+

5. soil_moisture_model.pth (246 MB)

   • Architecture: ResNet34
   • Task: Soil moisture level classification
   • Classes: 3 levels (Dry, Moderate, Wet)
   • Accuracy: 93%+

Usage

from huggingface_hub import hf_hub_download
import torch

# Download a model
model_path = hf_hub_download(
    repo_id="watermelon-elite/aaagriculture-models",
    filename="soil_moisture_model.pth"
)

# Load the model
checkpoint = torch.load(model_path, map_location='cpu')
# ... load into your model architecture

Source Code

Full application code: https://github.com/watermelon-elite/AAAgriculture

License

MIT License - Free to use for research and commercial purposes.

Citation

If you use these models, please cite:

@software{aaagriculture2026,
  author = {Your Name},
  title = {AAAgriculture: ML-Powered Precision Agriculture Platform},
  year = {2026},
  url = {https://github.com/watermelon-elite/AAAgriculture}
}