edena

🌱 Edena - Soil Classification Model

EfficientNet-B0 model fine-tuned for soil type classification from images.

Model Description

This model classifies soil images into 7 different types using transfer learning with EfficientNet-B0.

Classes: Alluvial_Soil, Arid_Soil, Black_Soil, Laterite_Soil, Mountain_Soil, Red_Soil, Yellow_Soil

Performance

• Test Accuracy: 88.27%
• Validation Accuracy: 82.58%
• Training Epochs: 10

Dataset

Trained on the Comprehensive Soil Classification Datasets from Kaggle.

• Total images: 1,188 (original)
• Split: 70% train / 15% validation / 15% test

Usage

import torch
from torchvision import models, transforms
from PIL import Image

# Load model
model = models.efficientnet_b0()
in_features = model.classifier[1].in_features
model.classifier[1] = torch.nn.Linear(in_features, 7)

checkpoint = torch.load('pytorch_model.bin', map_location='cpu')
model.load_state_dict(checkpoint['model_state_dict'])
model.eval()

# Prepare image
transform = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

img = Image.open('soil_image.jpg').convert('RGB')
img_tensor = transform(img).unsqueeze(0)

# Predict
with torch.no_grad():
    output = model(img_tensor)
    pred_idx = torch.argmax(output, dim=1).item()

classes = ['Alluvial_Soil', 'Arid_Soil', 'Black_Soil', 'Laterite_Soil', 'Mountain_Soil', 'Red_Soil', 'Yellow_Soil']
print(f"Predicted soil type: {classes[pred_idx]}")

Training Details

• Framework: PyTorch
• Base Model: EfficientNet-B0 (pretrained on ImageNet)
• Optimizer: Adam
• Learning Rate: 0.001
• Batch Size: 32

Model Architecture

Transfer learning approach:

1. Loaded EfficientNet-B0 pretrained on ImageNet
2. Froze all layers except the final classifier
3. Replaced classifier head with 7-class output
4. Fine-tuned on soil classification dataset

License

• Code: Apache 2.0
• Model weights: CC BY-NC-SA 4.0

Project

GitHub: MrLilian24/edena

Citation

If you use this model in your research, please cite:

@misc{edena2025,
  author = {Lavergne, Rémi},
  title = {Edena: Soil Classification using Deep Learning},
  year = {2025},
  publisher = {HuggingFace},
  url = {https://huggingface.co/MrLilian24/edena}
}